Integrate transcription in PeerTube

This commit is contained in:
Chocobozzz 2024-06-13 09:23:12 +02:00
parent ef14cf4a5c
commit 1bfb791e05
No known key found for this signature in database
GPG Key ID: 583A612D890159BE
172 changed files with 2674 additions and 945 deletions

View File

@ -71,6 +71,20 @@ jobs:
${{ runner.OS }}-fixtures-
${{ runner.OS }}-
- name: Cache PeerTube pip directory
uses: actions/cache@v4
with:
path: |
~/.cache/pip
key: ${{ runner.OS }}-${{ matrix.test_suite }}-pip-v1
- name: Cache Hugging Face models
uses: actions/cache@v4
with:
path: |
~/.cache/huggingface
key: ${{ runner.OS }}-${{ matrix.test_suite }}-hugging-face-v1
- name: Set env test variable (schedule)
if: github.event_name != 'schedule'
run: |

3
.gitignore vendored
View File

@ -12,8 +12,11 @@ yarn-error.log
/test4/
/test5/
/test6/
# Big fixtures generated/downloaded on-demand
/packages/tests/fixtures/video_high_bitrate_1080p.mp4
/packages/tests/fixtures/video_59fps.mp4
/packages/tests/fixtures/transcription/models-v1/
# Production
/storage

View File

@ -1,6 +1,7 @@
import {
RunnerJobLiveRTMPHLSTranscodingPayload,
RunnerJobStudioTranscodingPayload,
RunnerJobTranscriptionPayload,
RunnerJobVODAudioMergeTranscodingPayload,
RunnerJobVODHLSTranscodingPayload,
RunnerJobVODWebVideoTranscodingPayload
@ -9,23 +10,39 @@ import { logger } from '../../shared/index.js'
import { processAudioMergeTranscoding, processHLSTranscoding, ProcessOptions, processWebVideoTranscoding } from './shared/index.js'
import { ProcessLiveRTMPHLSTranscoding } from './shared/process-live.js'
import { processStudioTranscoding } from './shared/process-studio.js'
import { processVideoTranscription } from './shared/process-transcription.js'
export async function processJob (options: ProcessOptions) {
const { server, job } = options
logger.info(`[${server.url}] Processing job of type ${job.type}: ${job.uuid}`, { payload: job.payload })
if (job.type === 'vod-audio-merge-transcoding') {
switch (job.type) {
case 'vod-audio-merge-transcoding':
await processAudioMergeTranscoding(options as ProcessOptions<RunnerJobVODAudioMergeTranscodingPayload>)
} else if (job.type === 'vod-web-video-transcoding') {
break
case 'vod-web-video-transcoding':
await processWebVideoTranscoding(options as ProcessOptions<RunnerJobVODWebVideoTranscodingPayload>)
} else if (job.type === 'vod-hls-transcoding') {
break
case 'vod-hls-transcoding':
await processHLSTranscoding(options as ProcessOptions<RunnerJobVODHLSTranscodingPayload>)
} else if (job.type === 'live-rtmp-hls-transcoding') {
break
case 'live-rtmp-hls-transcoding':
await new ProcessLiveRTMPHLSTranscoding(options as ProcessOptions<RunnerJobLiveRTMPHLSTranscodingPayload>).process()
} else if (job.type === 'video-studio-transcoding') {
break
case 'video-studio-transcoding':
await processStudioTranscoding(options as ProcessOptions<RunnerJobStudioTranscodingPayload>)
} else {
break
case 'video-transcription':
await processVideoTranscription(options as ProcessOptions<RunnerJobTranscriptionPayload>)
break
default:
logger.error(`Unknown job ${job.type} to process`)
return
}

View File

@ -5,7 +5,7 @@ import { RunnerJob, RunnerJobPayload } from '@peertube/peertube-models'
import { buildUUID } from '@peertube/peertube-node-utils'
import { PeerTubeServer } from '@peertube/peertube-server-commands'
import { ConfigManager, downloadFile, logger } from '../../../shared/index.js'
import { getTranscodingLogger } from './transcoding-logger.js'
import { getWinstonLogger } from './winston-logger.js'
export type JobWithToken <T extends RunnerJobPayload = RunnerJobPayload> = RunnerJob<T> & { jobToken: string }
@ -101,6 +101,6 @@ function getCommonFFmpegOptions () {
available: getDefaultAvailableEncoders(),
encodersToTry: getDefaultEncodersToTry()
},
logger: getTranscodingLogger()
logger: getWinstonLogger()
}
}

View File

@ -1,3 +1,3 @@
export * from './common.js'
export * from './process-vod.js'
export * from './transcoding-logger.js'
export * from './winston-logger.js'

View File

@ -0,0 +1,79 @@
import { hasAudioStream } from '@peertube/peertube-ffmpeg'
import { RunnerJobTranscriptionPayload, TranscriptionSuccess } from '@peertube/peertube-models'
import { buildSUUID } from '@peertube/peertube-node-utils'
import { TranscriptionModel, WhisperBuiltinModel, transcriberFactory } from '@peertube/peertube-transcription'
import { remove } from 'fs-extra/esm'
import { join } from 'path'
import { ConfigManager } from '../../../shared/config-manager.js'
import { logger } from '../../../shared/index.js'
import { ProcessOptions, downloadInputFile, scheduleTranscodingProgress } from './common.js'
import { getWinstonLogger } from './winston-logger.js'
export async function processVideoTranscription (options: ProcessOptions<RunnerJobTranscriptionPayload>) {
const { server, job, runnerToken } = options
const config = ConfigManager.Instance.getConfig().transcription
const payload = job.payload
let inputPath: string
const updateProgressInterval = scheduleTranscodingProgress({
job,
server,
runnerToken,
progressGetter: () => undefined
})
const outputPath = join(ConfigManager.Instance.getTranscriptionDirectory(), buildSUUID())
const transcriber = transcriberFactory.createFromEngineName({
engineName: config.engine,
enginePath: config.enginePath,
logger: getWinstonLogger()
})
try {
logger.info(`Downloading input file ${payload.input.videoFileUrl} for transcription job ${job.jobToken}`)
inputPath = await downloadInputFile({ url: payload.input.videoFileUrl, runnerToken, job })
logger.info(`Downloaded input file ${payload.input.videoFileUrl} for job ${job.jobToken}. Running transcription.`)
if (await hasAudioStream(inputPath) !== true) {
await server.runnerJobs.error({
jobToken: job.jobToken,
jobUUID: job.uuid,
runnerToken,
message: 'This input file does not contain audio'
})
return
}
const transcriptFile = await transcriber.transcribe({
mediaFilePath: inputPath,
model: config.modelPath
? await TranscriptionModel.fromPath(config.modelPath)
: new WhisperBuiltinModel(config.model),
format: 'vtt',
transcriptDirectory: outputPath
})
const successBody: TranscriptionSuccess = {
inputLanguage: transcriptFile.language,
vttFile: transcriptFile.path
}
await server.runnerJobs.success({
jobToken: job.jobToken,
jobUUID: job.uuid,
runnerToken,
payload: successBody
})
} finally {
if (inputPath) await remove(inputPath)
if (outputPath) await remove(outputPath)
if (updateProgressInterval) clearInterval(updateProgressInterval)
}
}

View File

@ -1,19 +0,0 @@
import { LogFn } from 'pino'
import { logger } from '../../../shared/index.js'
export function getTranscodingLogger () {
return {
info: buildWinstonLogger(logger.info.bind(logger)),
debug: buildWinstonLogger(logger.debug.bind(logger)),
warn: buildWinstonLogger(logger.warn.bind(logger)),
error: buildWinstonLogger(logger.error.bind(logger))
}
}
function buildWinstonLogger (log: LogFn) {
return (arg1: string, arg2?: object) => {
if (arg2) return log(arg2, arg1)
return log(arg1)
}
}

View File

@ -0,0 +1,19 @@
import { LogFn } from 'pino'
import { logger } from '../../../shared/index.js'
export function getWinstonLogger () {
return {
info: buildLogLevelFn(logger.info.bind(logger)),
debug: buildLogLevelFn(logger.debug.bind(logger)),
warn: buildLogLevelFn(logger.warn.bind(logger)),
error: buildLogLevelFn(logger.error.bind(logger))
}
}
function buildLogLevelFn (log: LogFn) {
return (arg1: string, arg2?: object) => {
if (arg2) return log(arg2, arg1)
return log(arg1)
}
}

View File

@ -1,15 +1,16 @@
import {
RunnerJobLiveRTMPHLSTranscodingPayload,
RunnerJobPayload,
RunnerJobType,
RunnerJobStudioTranscodingPayload,
RunnerJobTranscriptionPayload,
RunnerJobType,
RunnerJobVODAudioMergeTranscodingPayload,
RunnerJobVODHLSTranscodingPayload,
RunnerJobVODWebVideoTranscodingPayload,
VideoStudioTaskPayload
} from '@peertube/peertube-models'
const supportedMatrix = {
const supportedMatrix: { [ id in RunnerJobType ]: (payload: RunnerJobPayload) => boolean } = {
'vod-web-video-transcoding': (_payload: RunnerJobVODWebVideoTranscodingPayload) => {
return true
},
@ -29,6 +30,9 @@ const supportedMatrix = {
if (!Array.isArray(tasks)) return false
return tasks.every(t => t && supported.has(t.name))
},
'video-transcription': (_payload: RunnerJobTranscriptionPayload) => {
return true
}
}

View File

@ -1,4 +1,5 @@
import { parse, stringify } from '@iarna/toml'
import { TranscriptionEngineName, WhisperBuiltinModelName } from '@peertube/peertube-transcription'
import envPaths from 'env-paths'
import { ensureDir, pathExists, remove } from 'fs-extra/esm'
import { readFile, writeFile } from 'fs/promises'
@ -24,6 +25,13 @@ type Config = {
runnerName: string
runnerDescription?: string
}[]
transcription: {
engine: TranscriptionEngineName
enginePath: string | null
model: WhisperBuiltinModelName
modelPath: string | null
}
}
export class ConfigManager {
@ -37,6 +45,12 @@ export class ConfigManager {
threads: 2,
nice: 20
},
transcription: {
engine: 'whisper-ctranslate2',
enginePath: null,
model: 'small',
modelPath: null
},
registeredInstances: []
}
@ -98,6 +112,10 @@ export class ConfigManager {
return join(paths.cache, this.id, 'transcoding')
}
getTranscriptionDirectory () {
return join(paths.cache, this.id, 'transcription')
}
getSocketDirectory () {
return join(paths.data, this.id)
}

View File

@ -359,7 +359,6 @@
</ng-container>
<ng-container formGroupName="storyboards">
<div class="form-group">
<my-peertube-checkbox
inputName="storyboardsEnabled" formControlName="enabled"
@ -370,7 +369,35 @@
</ng-container>
</my-peertube-checkbox>
</div>
</ng-container>
<ng-container formGroupName="videoTranscription">
<div class="form-group">
<my-peertube-checkbox
inputName="videoTranscriptionEnabled" formControlName="enabled"
i18n-labelText labelText="Enable video transcription"
>
<ng-container ngProjectAs="description">
<span i18n>Automatically create a subtitle file of uploaded/imported VOD videos</span>
</ng-container>
<ng-container ngProjectAs="extra">
<div class="form-group" formGroupName="remoteRunners" [ngClass]="getTranscriptionRunnerDisabledClass()">
<my-peertube-checkbox
inputName="videoTranscriptionRemoteRunnersEnabled" formControlName="enabled"
i18n-labelText labelText="Enable remote runners for transcription"
>
<ng-container ngProjectAs="description">
<span i18n>
Use <a routerLink="/admin/system/runners/runners-list">remote runners</a> to process transcription tasks.
Remote runners has to register on your instance first.
</span>
</ng-container>
</my-peertube-checkbox>
</div>
</ng-container>
</my-peertube-checkbox>
</div>
</ng-container>
</div>
</div>

View File

@ -137,6 +137,18 @@ export class EditBasicConfigurationComponent implements OnInit, OnChanges {
return { 'disabled-checkbox-extra': !this.isSearchIndexEnabled() }
}
// ---------------------------------------------------------------------------
isTranscriptionEnabled () {
return this.form.value['videoTranscription']['enabled'] === true
}
getTranscriptionRunnerDisabledClass () {
return { 'disabled-checkbox-extra': !this.isTranscriptionEnabled() }
}
// ---------------------------------------------------------------------------
isAutoFollowIndexEnabled () {
return this.form.value['followings']['instance']['autoFollowIndex']['enabled'] === true
}

View File

@ -267,6 +267,12 @@ export class EditCustomConfigComponent extends FormReactive implements OnInit {
enabled: null
}
},
videoTranscription: {
enabled: null,
remoteRunners: {
enabled: null
}
},
videoFile: {
update: {
enabled: null

View File

@ -1,7 +1,7 @@
import { DatePipe, NgClass, NgFor, NgIf } from '@angular/common'
import { Component, OnInit, ViewChild } from '@angular/core'
import { ActivatedRoute, Router, RouterLink } from '@angular/router'
import { AuthService, ConfirmService, Notifier, RestPagination, RestTable } from '@app/core'
import { AuthService, ConfirmService, Notifier, RestPagination, RestTable, ServerService } from '@app/core'
import { formatICU, getAbsoluteAPIUrl } from '@app/helpers'
import { VideoDetails } from '@app/shared/shared-main/video/video-details.model'
import { VideoFileTokenService } from '@app/shared/shared-main/video/video-file-token.service'
@ -30,6 +30,7 @@ import {
VideoActionsDropdownComponent
} from '../../../shared/shared-video-miniature/video-actions-dropdown.component'
import { VideoAdminService } from './video-admin.service'
import { VideoCaptionService } from '@app/shared/shared-main/video-caption/video-caption.service'
@Component({
selector: 'my-video-list',
@ -84,7 +85,8 @@ export class VideoListComponent extends RestTable <Video> implements OnInit {
removeFiles: true,
transcoding: true,
studio: true,
stats: true
stats: true,
generateTranscription: true
}
loading = true
@ -100,6 +102,8 @@ export class VideoListComponent extends RestTable <Video> implements OnInit {
private videoService: VideoService,
private videoAdminService: VideoAdminService,
private videoBlockService: VideoBlockService,
private videoCaptionService: VideoCaptionService,
private server: ServerService,
private videoFileTokenService: VideoFileTokenService
) {
super()
@ -109,6 +113,10 @@ export class VideoListComponent extends RestTable <Video> implements OnInit {
return this.auth.getUser()
}
get serverConfig () {
return this.server.getHTMLConfig()
}
ngOnInit () {
this.initialize()
@ -160,6 +168,14 @@ export class VideoListComponent extends RestTable <Video> implements OnInit {
isDisplayed: videos => videos.every(v => v.canRemoveFiles(this.authUser)),
iconName: 'delete'
}
],
[
{
label: $localize`Generate caption`,
handler: videos => this.generateCaption(videos),
isDisplayed: videos => videos.every(v => v.canGenerateTranscription(this.authUser, this.serverConfig.videoTranscription.enabled)),
iconName: 'video-lang'
}
]
]
}
@ -399,4 +415,15 @@ export class VideoListComponent extends RestTable <Video> implements OnInit {
error: err => this.notifier.error(err.message)
})
}
private generateCaption (videos: Video[]) {
this.videoCaptionService.generateCaption(videos.map(v => v.id))
.subscribe({
next: () => {
this.notifier.success($localize`Transcription jobs created.`)
},
error: err => this.notifier.error(err.message)
})
}
}

View File

@ -69,6 +69,7 @@ export class JobsComponent extends RestTable implements OnInit {
'video-redundancy',
'video-studio-edition',
'video-transcoding',
'video-transcription',
'videos-views-stats'
]

View File

@ -49,7 +49,8 @@ export class MyAccountNotificationPreferencesComponent implements OnInit {
abuseStateChange: $localize`One of your abuse reports has been accepted or rejected by moderators`,
newPeerTubeVersion: $localize`A new PeerTube version is available`,
newPluginVersion: $localize`One of your plugin/theme has a new available version`,
myVideoStudioEditionFinished: $localize`Video studio edition has finished`
myVideoStudioEditionFinished: $localize`Video studio edition has finished`,
myVideoTranscriptionGenerated: $localize`The transcription of your video has been generated`
}
this.notificationSettingGroups = [
{
@ -68,7 +69,8 @@ export class MyAccountNotificationPreferencesComponent implements OnInit {
'blacklistOnMyVideo',
'myVideoPublished',
'myVideoImportFinished',
'myVideoStudioEditionFinished'
'myVideoStudioEditionFinished',
'myVideoTranscriptionGenerated'
]
},

View File

@ -173,12 +173,8 @@
<ng-template ngbNavContent>
<div class="captions">
<div class="captions-header">
<button (click)="openAddCaptionModal()" class="peertube-create-button">
<my-global-icon iconName="add" aria-hidden="true"></my-global-icon>
<ng-container i18n>Add another caption</ng-container>
</button>
<div class="alert pt-alert-primary" *ngIf="displayTranscriptionInfo && isTranscriptionEnabled()" i18n>
A subtitle will be automatically generated from your video.
</div>
<div class="form-group" *ngFor="let videoCaption of videoCaptions">
@ -226,6 +222,13 @@
No captions for now.
</div>
<div class="mt-3 mb-3">
<button (click)="openAddCaptionModal()" class="peertube-create-button">
<my-global-icon iconName="add" aria-hidden="true"></my-global-icon>
<ng-container i18n>Add a caption</ng-container>
</button>
</div>
</div>
</ng-template>
</ng-container>

View File

@ -24,11 +24,6 @@ my-peertube-checkbox {
}
}
.captions-header {
text-align: end;
margin-bottom: 1rem;
}
.caption-entry {
display: flex;
height: 40px;

View File

@ -1,5 +1,16 @@
import { DatePipe, NgClass, NgFor, NgIf, NgTemplateOutlet } from '@angular/common'
import { ChangeDetectorRef, Component, EventEmitter, Input, NgZone, OnDestroy, OnInit, Output, ViewChild } from '@angular/core'
import {
ChangeDetectorRef,
Component,
EventEmitter,
Input,
NgZone,
OnDestroy,
OnInit,
Output,
ViewChild,
booleanAttribute
} from '@angular/core'
import { AbstractControl, FormArray, FormGroup, FormsModule, ReactiveFormsModule, Validators } from '@angular/forms'
import { HooksService, PluginService, ServerService } from '@app/core'
import { removeElementFromArray } from '@app/helpers'
@ -63,10 +74,10 @@ import { HelpComponent } from '../../../shared/shared-main/misc/help.component'
import { EmbedComponent } from '../../../shared/shared-main/video/embed.component'
import { LiveDocumentationLinkComponent } from '../../../shared/shared-video-live/live-documentation-link.component'
import { I18nPrimengCalendarService } from './i18n-primeng-calendar.service'
import { ThumbnailManagerComponent } from './thumbnail-manager/thumbnail-manager.component'
import { VideoCaptionAddModalComponent } from './video-caption-add-modal.component'
import { VideoCaptionEditModalContentComponent } from './video-caption-edit-modal-content/video-caption-edit-modal-content.component'
import { VideoEditType } from './video-edit.type'
import { ThumbnailManagerComponent } from './thumbnail-manager/thumbnail-manager.component'
type VideoLanguages = VideoConstant<string> & { group?: string }
type PluginField = {
@ -122,15 +133,17 @@ export class VideoEditComponent implements OnInit, OnDestroy {
@Input() publishedVideo: VideoDetails
@Input() userVideoChannels: SelectChannelItem[] = []
@Input() forbidScheduledPublication = true
@Input({ transform: booleanAttribute }) forbidScheduledPublication = true
@Input({ transform: booleanAttribute }) displayTranscriptionInfo = true
@Input() videoCaptions: VideoCaptionWithPathEdit[] = []
@Input() videoSource: VideoSource
@Input() videoChapters: VideoChapter[] = []
@Input() hideWaitTranscoding = false
@Input() updateVideoFileEnabled = false
@Input({ transform: booleanAttribute }) hideWaitTranscoding = false
@Input({ transform: booleanAttribute }) updateVideoFileEnabled = false
@Input() type: VideoEditType
@Input() liveVideo: LiveVideo
@ -405,6 +418,10 @@ export class VideoEditComponent implements OnInit, OnDestroy {
return !!this.form.value['originallyPublishedAt']
}
isTranscriptionEnabled () {
return this.serverConfig.videoTranscription.enabled
}
// ---------------------------------------------------------------------------
resetField (name: string) {

View File

@ -53,8 +53,8 @@
<form [hidden]="!isInUpdateForm" novalidate [formGroup]="form">
<my-video-edit
[form]="form" [formErrors]="formErrors" [videoCaptions]="videoCaptions"
[forbidScheduledPublication]="true" [hideWaitTranscoding]="true"
[validationMessages]="validationMessages" [userVideoChannels]="userVideoChannels" [liveVideo]="liveVideo"
forbidScheduledPublication="true" hideWaitTranscoding="true" displayTranscriptionInfo="false"
type="go-live"
></my-video-edit>

View File

@ -59,8 +59,9 @@
<!-- Hidden because we want to load the component -->
<form [hidden]="!hasImportedVideo" novalidate [formGroup]="form">
<my-video-edit
[form]="form" [formErrors]="formErrors" [videoCaptions]="videoCaptions" [forbidScheduledPublication]="true"
[form]="form" [formErrors]="formErrors" [videoCaptions]="videoCaptions"
[validationMessages]="validationMessages" [userVideoChannels]="userVideoChannels"
forbidScheduledPublication="true"
type="import-torrent"
></my-video-edit>

View File

@ -57,8 +57,9 @@
<form [hidden]="!hasImportedVideo" novalidate [formGroup]="form">
<my-video-edit
#videoEdit
[form]="form" [formErrors]="formErrors" [videoCaptions]="videoCaptions" [forbidScheduledPublication]="true"
[form]="form" [formErrors]="formErrors" [videoCaptions]="videoCaptions"
[validationMessages]="validationMessages" [userVideoChannels]="userVideoChannels"
forbidScheduledPublication="true"
type="import-url"
></my-video-edit>

View File

@ -67,7 +67,7 @@
<my-video-edit
[form]="form" [formErrors]="formErrors" [videoCaptions]="videoCaptions"
[validationMessages]="validationMessages" [userVideoChannels]="userVideoChannels"
[forbidScheduledPublication]="false"
forbidScheduledPublication="false"
type="upload"
></my-video-edit>

View File

@ -20,6 +20,7 @@
type="update" (pluginFieldsAdded)="hydratePluginFieldsFromVideo()"
[liveVideo]="liveVideo" [publishedVideo]="videoDetails"
[videoSource]="videoSource" [updateVideoFileEnabled]="isUpdateVideoFileEnabled()"
displayTranscriptionInfo="false"
(formBuilt)="onFormBuilt()"
>

View File

@ -34,6 +34,13 @@
</td>
</tr>
<tr>
<th i18n class="sub-label" scope="row">Automatic transcription</th>
<td>
<my-feature-boolean [value]="serverConfig.videoTranscription.enabled"></my-feature-boolean>
</td>
</tr>
<tr>
<th i18n class="sub-label" scope="row">Video uploads</th>
<td>

View File

@ -63,6 +63,10 @@
&.with-icon {
@include dropdown-with-icon-item;
.icon-video-lang {
top: 0;
}
}
a,

View File

@ -11,6 +11,7 @@ import {
UserNotificationType,
UserNotificationType_Type,
UserRight,
VideoConstant,
VideoInfo
} from '@peertube/peertube-models'
import { logger } from '@root-helpers/logger'
@ -90,6 +91,12 @@ export class UserNotification implements UserNotificationServer {
username: string
}
videoCaption?: {
id: number
language: VideoConstant<string>
video: VideoInfo
}
createdAt: string
updatedAt: string
@ -149,6 +156,8 @@ export class UserNotification implements UserNotificationServer {
this.peertube = hash.peertube
this.registration = hash.registration
this.videoCaption = hash.videoCaption
this.createdAt = hash.createdAt
this.updatedAt = hash.updatedAt
@ -250,6 +259,10 @@ export class UserNotification implements UserNotificationServer {
this.pluginQueryParams.pluginType = this.plugin.type + ''
break
case UserNotificationType.MY_VIDEO_TRANSCRIPTION_GENERATED:
this.videoUrl = this.buildVideoUrl(this.videoCaption.video)
break
case UserNotificationType.MY_VIDEO_STUDIO_EDITION_FINISHED:
this.videoUrl = this.buildVideoUrl(this.video)
break

View File

@ -1,15 +1,15 @@
import { Observable, of } from 'rxjs'
import { catchError, map, switchMap } from 'rxjs/operators'
import { HttpClient } from '@angular/common/http'
import { Injectable } from '@angular/core'
import { RestExtractor, ServerService } from '@app/core'
import { objectToFormData } from '@app/helpers'
import { peertubeTranslate, sortBy } from '@peertube/peertube-core-utils'
import { ResultList, VideoCaption } from '@peertube/peertube-models'
import { Observable, from, of } from 'rxjs'
import { catchError, concatMap, map, switchMap, toArray } from 'rxjs/operators'
import { environment } from '../../../../environments/environment'
import { VideoCaptionEdit } from './video-caption-edit.model'
import { VideoPasswordService } from '../video/video-password.service'
import { VideoService } from '../video/video.service'
import { VideoCaptionEdit } from './video-caption-edit.model'
@Injectable()
export class VideoCaptionService {
@ -74,4 +74,13 @@ export class VideoCaptionService {
getCaptionContent ({ captionPath }: Pick<VideoCaption, 'captionPath'>) {
return this.authHttp.get(environment.originServerUrl + captionPath, { responseType: 'text' })
}
generateCaption (videoIds: (number | string)[]) {
return from(videoIds)
.pipe(
concatMap(videoId => this.authHttp.post(`${VideoService.BASE_VIDEO_URL}/${videoId}/captions/generate`, {})),
toArray(),
catchError(err => this.restExtractor.handleError(err))
)
}
}

View File

@ -244,6 +244,10 @@ export class Video implements VideoServerModel {
this.isUpdatableBy(user)
}
canGenerateTranscription (user: AuthUser, transcriptionEnabled: boolean) {
return transcriptionEnabled && this.isLocal && user.hasRight(UserRight.UPDATE_ANY_VIDEO)
}
// ---------------------------------------------------------------------------
isOwner (user: AuthUser) {

View File

@ -11,6 +11,7 @@ import {
DropdownButtonSize,
DropdownDirection
} from '../shared-main/buttons/action-dropdown.component'
import { VideoCaptionService } from '../shared-main/video-caption/video-caption.service'
import { RedundancyService } from '../shared-main/video/redundancy.service'
import { VideoDetails } from '../shared-main/video/video-details.model'
import { Video } from '../shared-main/video/video.model'
@ -37,6 +38,7 @@ export type VideoActionsDisplayType = {
transcoding?: boolean
studio?: boolean
stats?: boolean
generateTranscription?: boolean
}
@Component({
@ -115,6 +117,7 @@ export class VideoActionsDropdownComponent implements OnChanges {
private videoBlocklistService: VideoBlockService,
private screenService: ScreenService,
private videoService: VideoService,
private videoCaptionService: VideoCaptionService,
private redundancyService: RedundancyService,
private serverService: ServerService
) { }
@ -206,6 +209,10 @@ export class VideoActionsDropdownComponent implements OnChanges {
return this.video.isLiveInfoAvailableBy(this.user)
}
canGenerateTranscription () {
return this.video.canGenerateTranscription(this.user, this.serverService.getHTMLConfig().videoTranscription.enabled)
}
isVideoDownloadableByAnonymous () {
return (
this.video &&
@ -338,7 +345,7 @@ export class VideoActionsDropdownComponent implements OnChanges {
this.videoService.runTranscoding({ videos: [ video ], type, askForForceTranscodingIfNeeded: true })
.subscribe({
next: () => {
this.notifier.success($localize`Transcoding jobs created for "${video.name}".`)
this.notifier.success($localize`Transcoding job created for "${video.name}".`)
this.transcodingCreated.emit()
},
@ -346,6 +353,17 @@ export class VideoActionsDropdownComponent implements OnChanges {
})
}
generateCaption (video: Video) {
this.videoCaptionService.generateCaption([ video.id ])
.subscribe({
next: () => {
this.notifier.success($localize`Transcription job created for "${video.name}".`)
},
error: err => this.notifier.error(err.message)
})
}
onVideoBlocked () {
this.videoBlocked.emit()
}
@ -466,6 +484,14 @@ export class VideoActionsDropdownComponent implements OnChanges {
iconName: 'delete'
}
],
[
{
label: $localize`Generate caption`,
handler: ({ video }) => this.generateCaption(video),
isDisplayed: () => this.displayOptions.generateTranscription && this.canGenerateTranscription(),
iconName: 'video-lang'
}
],
[ // actions regarding the account/its server
{
label: $localize`Mute account`,

View File

@ -239,6 +239,14 @@
}
</ng-container>
<ng-container *ngSwitchCase="22"> <!-- UserNotificationType.MY_VIDEO_TRANSCRIPTION_GENERATED -->
<my-global-icon iconName="video-lang" aria-hidden="true"></my-global-icon>
<div class="message" i18n>
<em>{{ notification.videoCaption.language.label }}</em> transcription of <a (click)="markAsRead(notification)" [routerLink]="notification.videoUrl">your video {{ notification.videoCaption.video.name }}</a> has been generated
</div>
</ng-container>
<ng-container *ngSwitchDefault>
<my-global-icon iconName="alert" aria-hidden="true"></my-global-icon>

View File

@ -716,6 +716,34 @@ video_studio:
remote_runners:
enabled: false
video_transcription:
# Enable automatic transcription of videos
enabled: false
# Choose engine for local transcription
# Supported: 'openai-whisper' or 'whisper-ctranslate2'
engine: 'whisper-ctranslate2'
# You can set a custom engine path for local transcription
# If not provided, PeerTube will try to automatically install it in the PeerTube bin directory
engine_path: null
# Choose engine model for local transcription
# Available for 'openai-whisper' and 'whisper-ctranslate2': 'tiny', 'base', 'small', 'medium' or 'large-v3'
model: 'small'
# Or specify the model path:
# * PyTorch model file path for 'openai-whisper'
# * CTranslate2 Whisper model directory path for 'whisper-ctranslate2'
# If not provided, PeerTube will automatically download the model
model_path: null
# Enable remote runners to transcribe videos
# If enabled, your instance won't transcribe the videos itself
# At least 1 remote runner must be configured to transcribe your videos
remote_runners:
enabled: false
video_file:
update:
# Add ability for users to replace the video file of an existing video

View File

@ -726,6 +726,34 @@ video_studio:
remote_runners:
enabled: false
video_transcription:
# Enable automatic transcription of videos
enabled: false
# Choose engine for local transcription
# Supported: 'openai-whisper' or 'whisper-ctranslate2'
engine: 'whisper-ctranslate2'
# You can set a custom engine path for local transcription
# If not provided, PeerTube will try to automatically install it in the PeerTube bin directory
engine_path: null
# Choose engine model for local transcription
# Available for 'openai-whisper' and 'whisper-ctranslate2': 'tiny', 'base', 'small', 'medium' or 'large-v3'
model: 'small'
# Or specify the model path:
# * PyTorch model file path for 'openai-whisper'
# * CTranslate2 Whisper model directory path for 'whisper-ctranslate2'
# If not provided, PeerTube will automatically download the model
model_path: null
# Enable remote runners to transcribe videos
# If enabled, your instance won't transcribe the videos itself
# At least 1 remote runner must be configured to transcribe your videos
remote_runners:
enabled: false
video_file:
update:
# Add ability for users to replace the video file of an existing video

View File

@ -166,3 +166,6 @@ open_telemetry:
search:
search_index:
url: 'https://search.joinpeertube.org/'
video_transcription:
model: 'tiny'

View File

@ -1,14 +1,13 @@
import { pick, promisify0 } from '@peertube/peertube-core-utils'
import { AvailableEncoders, EncoderOptionsBuilder, EncoderOptionsBuilderParams, EncoderProfile } from '@peertube/peertube-models'
import {
AvailableEncoders,
EncoderOptionsBuilder,
EncoderOptionsBuilderParams,
EncoderProfile,
SimpleLogger
} from '@peertube/peertube-models'
import ffmpeg, { FfmpegCommand } from 'fluent-ffmpeg'
type FFmpegLogger = {
info: (msg: string, obj?: object) => void
debug: (msg: string, obj?: object) => void
warn: (msg: string, obj?: object) => void
error: (msg: string, obj?: object) => void
}
export interface FFmpegCommandWrapperOptions {
availableEncoders?: AvailableEncoders
profile?: string
@ -17,7 +16,7 @@ export interface FFmpegCommandWrapperOptions {
tmpDirectory: string
threads: number
logger: FFmpegLogger
logger: SimpleLogger
lTags?: { tags: string[] }
updateJobProgress?: (progress?: number) => void
@ -35,7 +34,7 @@ export class FFmpegCommandWrapper {
private readonly tmpDirectory: string
private readonly threads: number
private readonly logger: FFmpegLogger
private readonly logger: SimpleLogger
private readonly lTags: { tags: string[] }
private readonly updateJobProgress: (progress?: number) => void

View File

@ -1,37 +0,0 @@
JiWER
=====
__JiWER__ CLI NodeJs wrapper.
> *JiWER is a python tool for computing the word-error-rate of ASR systems.*
> https://jitsi.github.io/jiwer/cli/
__JiWER__ serves as a reference implementation to calculate errors rates between 2 text files:
- WER (Word Error Rate)
- CER (Character Error Rate)
Build
-----
```sh
npm run build
```
Usage
-----
```typescript
const jiwerCLI = new JiwerClI('./reference.txt', './hypothesis.txt')
// WER as a percentage, ex: 0.03 -> 3%
console.log(await jiwerCLI.wer())
// CER as a percentage: 0.01 -> 1%
console.log(await jiwerCLI.cer())
// Detailed comparison report
console.log(await jiwerCLI.alignment())
```
Resources
---------
- https://jitsi.github.io/jiwer/
- https://github.com/rapidfuzz/RapidFuzz

View File

@ -1 +0,0 @@
jiwer==3.0.4

View File

@ -1 +0,0 @@
export * from './jiwer-cli.js'

View File

@ -1,2 +1,3 @@
export * from './file-storage.enum.js'
export * from './result-list.model.js'
export * from './simple-logger.model.js'

View File

@ -0,0 +1,6 @@
export type SimpleLogger = {
info: (msg: string, obj?: object) => void
debug: (msg: string, obj?: object) => void
warn: (msg: string, obj?: object) => void
error: (msg: string, obj?: object) => void
}

View File

@ -8,7 +8,8 @@ export type RunnerJobVODPayload =
export type RunnerJobPayload =
RunnerJobVODPayload |
RunnerJobLiveRTMPHLSTranscodingPayload |
RunnerJobStudioTranscodingPayload
RunnerJobStudioTranscodingPayload |
RunnerJobTranscriptionPayload
// ---------------------------------------------------------------------------
@ -54,6 +55,12 @@ export interface RunnerJobStudioTranscodingPayload {
tasks: VideoStudioTaskPayload[]
}
export interface RunnerJobTranscriptionPayload {
input: {
videoFileUrl: string
}
}
// ---------------------------------------------------------------------------
export function isAudioMergeTranscodingPayload (payload: RunnerJobPayload): payload is RunnerJobVODAudioMergeTranscodingPayload {

View File

@ -8,7 +8,8 @@ export type RunnerJobVODPrivatePayload =
export type RunnerJobPrivatePayload =
RunnerJobVODPrivatePayload |
RunnerJobLiveRTMPHLSTranscodingPrivatePayload |
RunnerJobVideoStudioTranscodingPrivatePayload
RunnerJobVideoStudioTranscodingPrivatePayload |
RunnerJobTranscriptionPrivatePayload
// ---------------------------------------------------------------------------
@ -45,3 +46,9 @@ export interface RunnerJobVideoStudioTranscodingPrivatePayload {
videoUUID: string
originalTasks: VideoStudioTaskPayload[]
}
// ---------------------------------------------------------------------------
export interface RunnerJobTranscriptionPrivatePayload {
videoUUID: string
}

View File

@ -12,7 +12,8 @@ export type RunnerJobSuccessPayload =
VODHLSTranscodingSuccess |
VODAudioMergeTranscodingSuccess |
LiveRTMPHLSTranscodingSuccess |
VideoStudioTranscodingSuccess
VideoStudioTranscodingSuccess |
TranscriptionSuccess
export interface VODWebVideoTranscodingSuccess {
videoFile: Blob | string
@ -35,6 +36,12 @@ export interface VideoStudioTranscodingSuccess {
videoFile: Blob | string
}
export interface TranscriptionSuccess {
inputLanguage: string
vttFile: Blob | string
}
export function isWebVideoOrAudioMergeTranscodingPayloadSuccess (
payload: RunnerJobSuccessPayload
): payload is VODHLSTranscodingSuccess | VODAudioMergeTranscodingSuccess {
@ -44,3 +51,7 @@ export function isWebVideoOrAudioMergeTranscodingPayloadSuccess (
export function isHLSTranscodingPayloadSuccess (payload: RunnerJobSuccessPayload): payload is VODHLSTranscodingSuccess {
return !!(payload as VODHLSTranscodingSuccess)?.resolutionPlaylistFile
}
export function isTranscriptionPayloadSuccess (payload: RunnerJobSuccessPayload): payload is TranscriptionSuccess {
return !!(payload as TranscriptionSuccess)?.vttFile
}

View File

@ -3,4 +3,5 @@ export type RunnerJobType =
'vod-hls-transcoding' |
'vod-audio-merge-transcoding' |
'live-rtmp-hls-transcoding' |
'video-studio-transcoding'
'video-studio-transcoding' |
'video-transcription'

View File

@ -179,6 +179,14 @@ export interface CustomConfig {
}
}
videoTranscription: {
enabled: boolean
remoteRunners: {
enabled: boolean
}
}
videoFile: {
update: {
enabled: boolean

View File

@ -33,6 +33,7 @@ export type JobType =
| 'generate-video-storyboard'
| 'create-user-export'
| 'import-user-archive'
| 'video-transcription'
export interface Job {
id: number | string
@ -101,11 +102,16 @@ export interface VideoImportYoutubeDLPayload {
type: VideoImportYoutubeDLPayloadType
videoImportId: number
generateTranscription: boolean
fileExt?: string
}
export interface VideoImportTorrentPayload {
type: VideoImportTorrentPayloadType
generateTranscription: boolean
videoImportId: number
}
@ -316,3 +322,9 @@ export interface CreateUserExportPayload {
export interface ImportUserArchivePayload {
userImportId: number
}
// ---------------------------------------------------------------------------
export interface VideoTranscriptionPayload {
videoUUID: string
}

View File

@ -346,6 +346,10 @@ export interface ServerConfig {
storyboards: {
enabled: boolean
}
videoTranscription: {
enabled: boolean
}
}
export type HTMLServerConfig = Omit<ServerConfig, 'signup'>

View File

@ -3,58 +3,59 @@ export const ServerErrorCode = {
* The simplest form of payload too large: when the file size is over the
* global file size limit
*/
MAX_FILE_SIZE_REACHED:'max_file_size_reached',
MAX_FILE_SIZE_REACHED: 'max_file_size_reached',
/**
* The payload is too large for the user quota set
*/
QUOTA_REACHED:'quota_reached',
QUOTA_REACHED: 'quota_reached',
/**
* Error yielded upon trying to access a video that is not federated, nor can
* be. This may be due to: remote videos on instances that are not followed by
* yours, and with your instance disallowing unknown instances being accessed.
*/
DOES_NOT_RESPECT_FOLLOW_CONSTRAINTS:'does_not_respect_follow_constraints',
DOES_NOT_RESPECT_FOLLOW_CONSTRAINTS: 'does_not_respect_follow_constraints',
LIVE_NOT_ENABLED:'live_not_enabled',
LIVE_NOT_ALLOWING_REPLAY:'live_not_allowing_replay',
LIVE_CONFLICTING_PERMANENT_AND_SAVE_REPLAY:'live_conflicting_permanent_and_save_replay',
LIVE_NOT_ENABLED: 'live_not_enabled',
LIVE_NOT_ALLOWING_REPLAY: 'live_not_allowing_replay',
LIVE_CONFLICTING_PERMANENT_AND_SAVE_REPLAY: 'live_conflicting_permanent_and_save_replay',
/**
* Pretty self-explanatory: the set maximum number of simultaneous lives was
* reached, and this error is typically there to inform the user trying to
* broadcast one.
*/
MAX_INSTANCE_LIVES_LIMIT_REACHED:'max_instance_lives_limit_reached',
MAX_INSTANCE_LIVES_LIMIT_REACHED: 'max_instance_lives_limit_reached',
/**
* Pretty self-explanatory: the set maximum number of simultaneous lives FOR
* THIS USER was reached, and this error is typically there to inform the user
* trying to broadcast one.
*/
MAX_USER_LIVES_LIMIT_REACHED:'max_user_lives_limit_reached',
MAX_USER_LIVES_LIMIT_REACHED: 'max_user_lives_limit_reached',
/**
* A torrent should have at most one correct video file. Any more and we will
* not be able to choose automatically.
*/
INCORRECT_FILES_IN_TORRENT:'incorrect_files_in_torrent',
INCORRECT_FILES_IN_TORRENT: 'incorrect_files_in_torrent',
COMMENT_NOT_ASSOCIATED_TO_VIDEO:'comment_not_associated_to_video',
COMMENT_NOT_ASSOCIATED_TO_VIDEO: 'comment_not_associated_to_video',
MISSING_TWO_FACTOR:'missing_two_factor',
INVALID_TWO_FACTOR:'invalid_two_factor',
MISSING_TWO_FACTOR: 'missing_two_factor',
INVALID_TWO_FACTOR: 'invalid_two_factor',
ACCOUNT_WAITING_FOR_APPROVAL:'account_waiting_for_approval',
ACCOUNT_APPROVAL_REJECTED:'account_approval_rejected',
ACCOUNT_WAITING_FOR_APPROVAL: 'account_waiting_for_approval',
ACCOUNT_APPROVAL_REJECTED: 'account_approval_rejected',
RUNNER_JOB_NOT_IN_PROCESSING_STATE:'runner_job_not_in_processing_state',
RUNNER_JOB_NOT_IN_PENDING_STATE:'runner_job_not_in_pending_state',
UNKNOWN_RUNNER_TOKEN:'unknown_runner_token',
RUNNER_JOB_NOT_IN_PROCESSING_STATE: 'runner_job_not_in_processing_state',
RUNNER_JOB_NOT_IN_PENDING_STATE: 'runner_job_not_in_pending_state',
UNKNOWN_RUNNER_TOKEN: 'unknown_runner_token',
VIDEO_REQUIRES_PASSWORD:'video_requires_password',
INCORRECT_VIDEO_PASSWORD:'incorrect_video_password',
VIDEO_REQUIRES_PASSWORD: 'video_requires_password',
INCORRECT_VIDEO_PASSWORD: 'incorrect_video_password',
VIDEO_ALREADY_BEING_TRANSCODED:'video_already_being_transcoded',
VIDEO_ALREADY_BEING_TRANSCODED: 'video_already_being_transcoded',
VIDEO_ALREADY_BEING_TRANSCRIBED: 'video_already_being_transcribed',
MAX_USER_VIDEO_QUOTA_EXCEEDED_FOR_USER_EXPORT: 'max_user_video_quota_exceeded_for_user_export'
} as const

View File

@ -31,4 +31,6 @@ export interface UserNotificationSetting {
newPluginVersion: UserNotificationSettingValueType
myVideoStudioEditionFinished: UserNotificationSettingValueType
myVideoTranscriptionGenerated: UserNotificationSettingValueType
}

View File

@ -1,6 +1,7 @@
import { FollowState } from '../actors/index.js'
import { AbuseStateType } from '../moderation/index.js'
import { PluginType_Type } from '../plugins/index.js'
import { VideoConstant } from '../videos/video-constant.model.js'
export const UserNotificationType = {
NEW_VIDEO_FROM_SUBSCRIPTION: 1,
@ -36,7 +37,9 @@ export const UserNotificationType = {
NEW_USER_REGISTRATION_REQUEST: 20,
NEW_LIVE_FROM_SUBSCRIPTION: 21
NEW_LIVE_FROM_SUBSCRIPTION: 21,
MY_VIDEO_TRANSCRIPTION_GENERATED: 22
} as const
export type UserNotificationType_Type = typeof UserNotificationType[keyof typeof UserNotificationType]
@ -138,6 +141,12 @@ export interface UserNotification {
username: string
}
videoCaption?: {
id: number
language: VideoConstant<string>
video: VideoInfo
}
createdAt: string
updatedAt: string
}

View File

@ -1,2 +1,3 @@
export * from './video-caption.model.js'
export * from './video-caption-generate.model.js'
export * from './video-caption-update.model.js'
export * from './video-caption.model.js'

View File

@ -0,0 +1,3 @@
export interface VideoCaptionGenerate {
forceTranscription?: boolean // Default false
}

View File

@ -5,5 +5,8 @@ export interface VideoImportCreate extends VideoUpdate {
magnetUri?: string
torrentfile?: Blob
// Default is true if the feature is enabled by the instance admin
generateTranscription?: boolean
channelId: number // Required
}

View File

@ -27,4 +27,7 @@ export interface VideoCreate {
thumbnailfile?: Blob | string
previewfile?: Blob | string
// Default is true if the feature is enabled by the instance admin
generateTranscription?: boolean
}

View File

@ -5,9 +5,6 @@ import {
AcceptRunnerJobResult,
ErrorRunnerJobBody,
HttpStatusCode,
isHLSTranscodingPayloadSuccess,
isLiveRTMPHLSTranscodingUpdatePayload,
isWebVideoOrAudioMergeTranscodingPayloadSuccess,
ListRunnerJobsQuery,
RequestRunnerJobBody,
RequestRunnerJobResult,
@ -22,8 +19,13 @@ import {
RunnerJobType,
RunnerJobUpdateBody,
RunnerJobVODPayload,
TranscriptionSuccess,
VODHLSTranscodingSuccess,
VODWebVideoTranscodingSuccess
VODWebVideoTranscodingSuccess,
isHLSTranscodingPayloadSuccess,
isLiveRTMPHLSTranscodingUpdatePayload,
isTranscriptionPayloadSuccess,
isWebVideoOrAudioMergeTranscodingPayloadSuccess
} from '@peertube/peertube-models'
import { unwrapBody } from '../requests/index.js'
import { waitJobs } from '../server/jobs.js'
@ -196,6 +198,12 @@ export class RunnerJobsCommand extends AbstractCommand {
payloadWithoutFiles = omit(payloadWithoutFiles as VODHLSTranscodingSuccess, [ 'resolutionPlaylistFile' ])
}
if (isTranscriptionPayloadSuccess(payload) && payload.vttFile) {
attaches[`payload[vttFile]`] = payload.vttFile
payloadWithoutFiles = omit(payloadWithoutFiles as TranscriptionSuccess, [ 'vttFile' ])
}
return this.postUploadRequest({
...options,

View File

@ -355,6 +355,29 @@ export class ConfigCommand extends AbstractCommand {
// ---------------------------------------------------------------------------
enableTranscription ({ remote = false }: { remote?: boolean } = {}) {
return this.setTranscriptionEnabled(true, remote)
}
disableTranscription () {
return this.setTranscriptionEnabled(false, false)
}
private setTranscriptionEnabled (enabled: boolean, remoteEnabled: boolean) {
return this.updateExistingConfig({
newConfig: {
videoTranscription: {
enabled,
remoteRunners: {
enabled: remoteEnabled
}
}
}
})
}
// ---------------------------------------------------------------------------
getConfig (options: OverrideCommandOptions = {}) {
const path = '/api/v1/config'

View File

@ -1,7 +1,10 @@
/* eslint-disable @typescript-eslint/no-unused-expressions,@typescript-eslint/no-floating-promises */
import { isAbsolute } from 'path'
import { HttpStatusCode, HttpStatusCodeType } from '@peertube/peertube-models'
import { buildAbsoluteFixturePath, getFileSize } from '@peertube/peertube-node-utils'
import { expect } from 'chai'
import got, { Response as GotResponse } from 'got'
import { isAbsolute } from 'path'
import {
makeDeleteRequest,
makeGetRequest,
@ -11,12 +14,9 @@ import {
unwrapBody,
unwrapText
} from '../requests/requests.js'
import { expect } from 'chai'
import got, { Response as GotResponse } from 'got'
import { HttpStatusCode, HttpStatusCodeType } from '@peertube/peertube-models'
import type { PeerTubeServer } from '../server/server.js'
import { createReadStream } from 'fs'
import type { PeerTubeServer } from '../server/server.js'
export interface OverrideCommandOptions {
token?: string

View File

@ -1,3 +1,4 @@
import { pick } from '@peertube/peertube-core-utils'
import { HttpStatusCode, ResultList, VideoCaption } from '@peertube/peertube-models'
import { buildAbsoluteFixturePath } from '@peertube/peertube-node-utils'
import { AbstractCommand, OverrideCommandOptions } from '../shared/index.js'
@ -32,6 +33,23 @@ export class CaptionsCommand extends AbstractCommand {
})
}
runGenerate (options: OverrideCommandOptions & {
videoId: string | number
forceTranscription?: boolean
}) {
const { videoId } = options
const path = '/api/v1/videos/' + videoId + '/captions/generate'
return this.postBodyRequest({
...options,
path,
fields: pick(options, [ 'forceTranscription' ]),
implicitToken: true,
defaultExpectedStatus: HttpStatusCode.NO_CONTENT_204
})
}
list (options: OverrideCommandOptions & {
videoId: string | number
videoPassword?: string

View File

@ -4,12 +4,19 @@ import { AbstractCommand, OverrideCommandOptions } from '../shared/index.js'
export class VideoImportsCommand extends AbstractCommand {
importVideo (options: OverrideCommandOptions & {
attributes: (VideoImportCreate | { torrentfile?: string, previewfile?: string, thumbnailfile?: string })
async importVideo (options: OverrideCommandOptions & {
attributes: (Partial<VideoImportCreate> | { torrentfile?: string, previewfile?: string, thumbnailfile?: string })
}) {
const { attributes } = options
const path = '/api/v1/videos/imports'
let defaultChannelId = 1
try {
const { videoChannels } = await this.server.users.getMyInfo({ token: options.token })
defaultChannelId = videoChannels[0].id
} catch (e) { /* empty */ }
let attaches: any = {}
if (attributes.torrentfile) attaches = { torrentfile: attributes.torrentfile }
if (attributes.thumbnailfile) attaches = { thumbnailfile: attributes.thumbnailfile }
@ -20,7 +27,11 @@ export class VideoImportsCommand extends AbstractCommand {
path,
attaches,
fields: options.attributes,
fields: {
channelId: defaultChannelId,
...options.attributes
},
implicitToken: true,
defaultExpectedStatus: HttpStatusCode.OK_200
}))

View File

@ -6,7 +6,8 @@
"devDependencies": {},
"scripts": {
"build": "tsc",
"watch": "tsc -w"
"watch": "tsc -w",
"install-dependencies:transcription": "pip install -r ./requirements.txt ../transcription-devtools/requirements.txt"
},
"dependencies": {}
}

View File

@ -0,0 +1,2 @@
whisper-ctranslate2
openai-whisper

View File

@ -42,6 +42,7 @@ import './video-source.js'
import './video-storyboards.js'
import './video-studio.js'
import './video-token.js'
import './video-transcription.js'
import './videos-common-filters.js'
import './videos-history.js'
import './videos-overviews.js'

View File

@ -171,6 +171,7 @@ describe('Test user notifications API validators', function () {
abuseStateChange: UserNotificationSettingValue.WEB,
newPeerTubeVersion: UserNotificationSettingValue.WEB,
myVideoStudioEditionFinished: UserNotificationSettingValue.WEB,
myVideoTranscriptionGenerated: UserNotificationSettingValue.WEB,
newPluginVersion: UserNotificationSettingValue.WEB
}

View File

@ -0,0 +1,106 @@
/* eslint-disable @typescript-eslint/no-unused-expressions,@typescript-eslint/require-await */
import { HttpStatusCode, UserRole } from '@peertube/peertube-models'
import {
PeerTubeServer,
cleanupTests,
createMultipleServers,
doubleFollow,
setAccessTokensToServers,
waitJobs
} from '@peertube/peertube-server-commands'
describe('Test video transcription API validator', function () {
let servers: PeerTubeServer[]
let userToken: string
let anotherUserToken: string
let remoteId: string
let validId: string
// ---------------------------------------------------------------
before(async function () {
this.timeout(240000)
servers = await createMultipleServers(2)
await setAccessTokensToServers(servers)
await doubleFollow(servers[0], servers[1])
userToken = await servers[0].users.generateUserAndToken('user', UserRole.USER)
anotherUserToken = await servers[0].users.generateUserAndToken('user2', UserRole.USER)
{
const { uuid } = await servers[1].videos.quickUpload({ name: 'remote video' })
remoteId = uuid
}
{
const { uuid } = await servers[0].videos.quickUpload({ name: 'both 1', token: userToken })
validId = uuid
}
await waitJobs(servers)
await servers[0].config.enableTranscription()
})
it('Should not run transcription of an unknown video', async function () {
await servers[0].captions.runGenerate({ videoId: 404, expectedStatus: HttpStatusCode.NOT_FOUND_404 })
})
it('Should not run transcription of a remote video', async function () {
await servers[0].captions.runGenerate({ videoId: remoteId, expectedStatus: HttpStatusCode.BAD_REQUEST_400 })
})
it('Should not run transcription by a owner/moderator user', async function () {
await servers[0].captions.runGenerate({ videoId: validId, token: anotherUserToken, expectedStatus: HttpStatusCode.FORBIDDEN_403 })
})
it('Should not run transcription if a caption file already exists', async function () {
await servers[0].captions.add({
language: 'en',
videoId: validId,
fixture: 'subtitle-good1.vtt'
})
await servers[0].captions.runGenerate({ videoId: validId, expectedStatus: HttpStatusCode.BAD_REQUEST_400 })
await servers[0].captions.delete({ language: 'en', videoId: validId })
})
it('Should not run transcription if the instance disabled it', async function () {
await servers[0].config.disableTranscription()
await servers[0].captions.runGenerate({ videoId: validId, expectedStatus: HttpStatusCode.BAD_REQUEST_400 })
await servers[0].config.enableTranscription()
})
it('Should succeed to run transcription', async function () {
await servers[0].captions.runGenerate({ videoId: validId, token: userToken })
})
it('Should fail to run transcription twice', async function () {
await servers[0].captions.runGenerate({ videoId: validId, token: userToken, expectedStatus: HttpStatusCode.CONFLICT_409 })
})
it('Should fail to run transcription twice with a non-admin user with the forceTranscription boolean', async function () {
await servers[0].captions.runGenerate({
videoId: validId,
token: userToken,
forceTranscription: true,
expectedStatus: HttpStatusCode.FORBIDDEN_403
})
})
it('Should succeed to run transcription twice with the forceTranscription boolean', async function () {
await servers[0].captions.runGenerate({ videoId: validId, forceTranscription: true })
})
after(async function () {
await cleanupTests(servers)
})
})

View File

@ -0,0 +1,81 @@
/* eslint-disable @typescript-eslint/no-unused-expressions,@typescript-eslint/require-await */
import { UserNotification } from '@peertube/peertube-models'
import { PeerTubeServer, cleanupTests, waitJobs } from '@peertube/peertube-server-commands'
import { MockSmtpServer } from '@tests/shared/mock-servers/mock-email.js'
import {
CheckerBaseParams,
checkMyVideoTranscriptionGenerated,
prepareNotificationsTest
} from '@tests/shared/notifications.js'
import { join } from 'path'
describe('Test caption notifications', function () {
let servers: PeerTubeServer[] = []
let userNotifications: UserNotification[] = []
let emails: object[] = []
let userAccessToken: string
before(async function () {
this.timeout(120000)
const res = await prepareNotificationsTest(1)
emails = res.emails
userAccessToken = res.userAccessToken
servers = res.servers
userNotifications = res.userNotifications
})
describe('Transcription of my video generated is published', function () {
const language = { id: 'en', label: 'English' }
let baseParams: CheckerBaseParams
before(() => {
baseParams = {
server: servers[0],
emails,
socketNotifications: userNotifications,
token: userAccessToken
}
})
async function uploadAndWait () {
const { uuid } = await servers[0].videos.upload({
token: userAccessToken,
attributes: {
name: 'video',
fixture: join('transcription', 'videos', 'the_last_man_on_earth.mp4'),
language: undefined
}
})
await waitJobs(servers)
return servers[0].videos.get({ id: uuid })
}
it('Should not send a notification if transcription is not enabled', async function () {
this.timeout(50000)
const { name, shortUUID } = await uploadAndWait()
await checkMyVideoTranscriptionGenerated({ ...baseParams, videoName: name, shortUUID, language, checkType: 'absence' })
})
it('Should send a notification transcription is enabled', async function () {
this.timeout(240000)
await servers[0].config.enableTranscription()
const { name, shortUUID } = await uploadAndWait()
await checkMyVideoTranscriptionGenerated({ ...baseParams, videoName: name, shortUUID, language, checkType: 'presence' })
})
})
after(async function () {
MockSmtpServer.Instance.kill()
await cleanupTests(servers)
})
})

View File

@ -1,4 +1,5 @@
import './admin-notifications.js'
import './captions-notifications.js'
import './comments-notifications.js'
import './moderation-notifications.js'
import './notifications-api.js'

View File

@ -2,4 +2,5 @@ export * from './runner-common.js'
export * from './runner-live-transcoding.js'
export * from './runner-socket.js'
export * from './runner-studio-transcoding.js'
export * from './runner-transcription.js'
export * from './runner-vod-transcoding.js'

View File

@ -0,0 +1,109 @@
/* eslint-disable @typescript-eslint/no-unused-expressions,@typescript-eslint/require-await */
import {
RunnerJobTranscriptionPayload,
TranscriptionSuccess
} from '@peertube/peertube-models'
import {
PeerTubeServer,
cleanupTests,
createMultipleServers,
doubleFollow,
setAccessTokensToServers,
setDefaultVideoChannel,
waitJobs
} from '@peertube/peertube-server-commands'
import { checkPersistentTmpIsEmpty } from '@tests/shared/directories.js'
import { expect } from 'chai'
describe('Test runner transcription', function () {
let servers: PeerTubeServer[] = []
let runnerToken: string
before(async function () {
this.timeout(120_000)
servers = await createMultipleServers(2)
await setAccessTokensToServers(servers)
await setDefaultVideoChannel(servers)
await doubleFollow(servers[0], servers[1])
await servers[0].config.enableTranscription({ remote: true })
runnerToken = await servers[0].runners.autoRegisterRunner()
})
async function upload () {
const { uuid } = await servers[0].videos.upload({ attributes: { name: 'video', language: undefined } })
const { availableJobs } = await servers[0].runnerJobs.request({ runnerToken })
expect(availableJobs).to.have.lengthOf(1)
const jobUUID = availableJobs[0].uuid
const { job } = await servers[0].runnerJobs.accept<RunnerJobTranscriptionPayload>({ runnerToken, jobUUID })
return { uuid, job }
}
it('Should execute a remote transcription job', async function () {
this.timeout(240_000)
const { uuid, job } = await upload()
expect(job.type === 'video-transcription')
expect(job.payload.input.videoFileUrl).to.exist
// Check video input file
{
await servers[0].runnerJobs.getJobFile({ url: job.payload.input.videoFileUrl, jobToken: job.jobToken, runnerToken })
}
const payload: TranscriptionSuccess = {
inputLanguage: 'ar',
vttFile: 'subtitle-good1.vtt'
}
await servers[0].runnerJobs.success({ runnerToken, jobUUID: job.uuid, jobToken: job.jobToken, payload })
await waitJobs(servers)
for (const server of servers) {
const video = await server.videos.get({ id: uuid })
expect(video.language.id).to.equal('ar')
const captions = await server.captions.list({ videoId: uuid })
expect(captions)
}
await checkPersistentTmpIsEmpty(servers[0])
})
it('Should not assign caption/language with an unknown inputLanguage', async function () {
this.timeout(240_000)
const { uuid, job } = await upload()
const payload: TranscriptionSuccess = {
inputLanguage: 'toto',
vttFile: 'subtitle-good1.vtt'
}
await servers[0].runnerJobs.success({ runnerToken, jobUUID: job.uuid, jobToken: job.jobToken, payload })
await waitJobs(servers)
for (const server of servers) {
const video = await server.videos.get({ id: uuid })
expect(video.language.id).to.be.null
const { total, data } = await server.captions.list({ videoId: uuid })
expect(total).to.equal(0)
expect(data).to.have.lengthOf(0)
}
})
after(async function () {
await cleanupTests(servers)
})
})

View File

@ -108,6 +108,9 @@ function checkInitialConfig (server: PeerTubeServer, data: CustomConfig) {
expect(data.videoStudio.enabled).to.be.false
expect(data.videoStudio.remoteRunners.enabled).to.be.false
expect(data.videoTranscription.enabled).to.be.false
expect(data.videoTranscription.remoteRunners.enabled).to.be.false
expect(data.videoFile.update.enabled).to.be.false
expect(data.import.videos.concurrency).to.equal(2)
@ -292,6 +295,12 @@ function buildNewCustomConfig (server: PeerTubeServer): CustomConfig {
enabled: true
}
},
videoTranscription: {
enabled: true,
remoteRunners: {
enabled: true
}
},
videoFile: {
update: {
enabled: true

View File

@ -1,24 +1,25 @@
import './channel-import-videos.js'
import './multiple-servers.js'
import './resumable-upload.js'
import './single-server.js'
import './video-captions.js'
import './video-change-ownership.js'
import './video-channel-syncs.js'
import './video-channels.js'
import './video-chapters.js'
import './channel-import-videos.js'
import './video-channel-syncs.js'
import './video-comments.js'
import './video-description.js'
import './video-files.js'
import './video-imports.js'
import './video-nsfw.js'
import './video-playlists.js'
import './video-playlist-thumbnails.js'
import './video-source.js'
import './video-playlists.js'
import './video-privacy.js'
import './video-schedule-update.js'
import './video-source.js'
import './video-static-file-privacy.js'
import './video-storyboard.js'
import './video-transcription.js'
import './videos-common-filters.js'
import './videos-history.js'
import './videos-overview.js'
import './video-static-file-privacy.js'
import './video-storyboard.js'

View File

@ -0,0 +1,145 @@
/* eslint-disable @typescript-eslint/no-unused-expressions,@typescript-eslint/require-await */
import { VideoPrivacy } from '@peertube/peertube-models'
import {
PeerTubeServer,
cleanupTests,
createMultipleServers,
doubleFollow,
sendRTMPStream,
setAccessTokensToServers,
setDefaultVideoChannel,
stopFfmpeg,
waitJobs
} from '@peertube/peertube-server-commands'
import { FIXTURE_URLS } from '@tests/shared/fixture-urls.js'
import { checkCaption, checkLanguage, checkNoCaption, uploadForTranscription } from '@tests/shared/transcription.js'
describe('Test video transcription', function () {
let servers: PeerTubeServer[]
before(async function () {
this.timeout(60000)
servers = await createMultipleServers(2)
await setAccessTokensToServers(servers)
await setDefaultVideoChannel(servers)
await doubleFollow(servers[0], servers[1])
await waitJobs(servers)
await waitJobs(servers)
})
// ---------------------------------------------------------------------------
it('Should generate a transcription on request', async function () {
this.timeout(360000)
await servers[0].config.disableTranscription()
const uuid = await uploadForTranscription(servers[0])
await waitJobs(servers)
await checkLanguage(servers, uuid, null)
await servers[0].config.enableTranscription()
await servers[0].captions.runGenerate({ videoId: uuid })
await waitJobs(servers)
await checkLanguage(servers, uuid, 'en')
await checkCaption(servers, uuid)
})
it('Should run transcription on upload by default', async function () {
this.timeout(360000)
const uuid = await uploadForTranscription(servers[0])
await waitJobs(servers)
await checkCaption(servers, uuid)
await checkLanguage(servers, uuid, 'en')
})
it('Should run transcription on import by default', async function () {
this.timeout(360000)
const { video } = await servers[0].videoImports.importVideo({
attributes: {
privacy: VideoPrivacy.PUBLIC,
targetUrl: FIXTURE_URLS.transcriptionVideo,
language: undefined
}
})
await waitJobs(servers)
await checkCaption(servers, video.uuid)
await checkLanguage(servers, video.uuid, 'en')
})
it('Should run transcription when live ended', async function () {
this.timeout(360000)
await servers[0].config.enableMinimumTranscoding()
await servers[0].config.enableLive({ allowReplay: true, transcoding: true, resolutions: 'min' })
const { live, video } = await servers[0].live.quickCreate({
saveReplay: true,
permanentLive: false,
privacy: VideoPrivacy.PUBLIC
})
const ffmpegCommand = sendRTMPStream({ rtmpBaseUrl: live.rtmpUrl, streamKey: live.streamKey })
await servers[0].live.waitUntilPublished({ videoId: video.id })
await stopFfmpeg(ffmpegCommand)
await servers[0].live.waitUntilReplacedByReplay({ videoId: video.id })
await waitJobs(servers)
await checkCaption(servers, video.uuid, 'WEBVTT\n\n00:')
await checkLanguage(servers, video.uuid, 'en')
await servers[0].config.enableLive({ allowReplay: false })
await servers[0].config.disableTranscoding()
})
it('Should not run transcription if disabled by user', async function () {
this.timeout(120000)
{
const uuid = await uploadForTranscription(servers[0], { generateTranscription: false })
await waitJobs(servers)
await checkNoCaption(servers, uuid)
await checkLanguage(servers, uuid, null)
}
{
const { video } = await servers[0].videoImports.importVideo({
attributes: {
privacy: VideoPrivacy.PUBLIC,
targetUrl: FIXTURE_URLS.transcriptionVideo,
generateTranscription: false
}
})
await waitJobs(servers)
await checkNoCaption(servers, video.uuid)
await checkLanguage(servers, video.uuid, null)
}
})
it('Should not run a transcription if the video does not contain audio', async function () {
this.timeout(120000)
const uuid = await uploadForTranscription(servers[0], { generateTranscription: false })
await waitJobs(servers)
await checkNoCaption(servers, uuid)
await checkLanguage(servers, uuid, null)
})
after(async function () {
await cleanupTests(servers)
})
})

View File

@ -3,7 +3,7 @@ import { buildAbsoluteFixturePath } from '@peertube/peertube-node-utils'
import { join } from 'path'
import { mkdir, rm, writeFile } from 'node:fs/promises'
import { expect } from 'chai'
import { JiwerClI } from '@peertube/peertube-jiwer'
import { JiwerClI } from '@peertube/peertube-transcription-devtools'
describe('Jiwer CLI', function () {
const transcriptDirectory = buildAbsoluteFixturePath('transcription/transcript-evaluator')

View File

@ -2,4 +2,5 @@ export * from './client-cli.js'
export * from './live-transcoding.js'
export * from './replace-file.js'
export * from './studio-transcoding.js'
export * from './video-transcription.js'
export * from './vod-transcoding.js'

View File

@ -194,7 +194,7 @@ describe('Test Live transcoding in peertube-runner program', function () {
describe('Check cleanup', function () {
it('Should have an empty cache directory', async function () {
await checkPeerTubeRunnerCacheIsEmpty(peertubeRunner)
await checkPeerTubeRunnerCacheIsEmpty(peertubeRunner, 'transcoding')
})
})

View File

@ -111,7 +111,7 @@ describe('Test studio transcoding in peertube-runner program', function () {
describe('Check cleanup', function () {
it('Should have an empty cache directory', async function () {
await checkPeerTubeRunnerCacheIsEmpty(peertubeRunner)
await checkPeerTubeRunnerCacheIsEmpty(peertubeRunner, 'transcoding')
})
})

View File

@ -0,0 +1,89 @@
/* eslint-disable @typescript-eslint/no-unused-expressions,@typescript-eslint/require-await */
import { wait } from '@peertube/peertube-core-utils'
import { RunnerJobState } from '@peertube/peertube-models'
import {
PeerTubeServer,
cleanupTests,
createMultipleServers,
doubleFollow,
setAccessTokensToServers,
setDefaultVideoChannel,
waitJobs
} from '@peertube/peertube-server-commands'
import { checkPeerTubeRunnerCacheIsEmpty } from '@tests/shared/directories.js'
import { PeerTubeRunnerProcess } from '@tests/shared/peertube-runner-process.js'
import { checkCaption, checkLanguage, checkNoCaption, uploadForTranscription } from '@tests/shared/transcription.js'
describe('Test transcription in peertube-runner program', function () {
let servers: PeerTubeServer[] = []
let peertubeRunner: PeerTubeRunnerProcess
before(async function () {
this.timeout(120_000)
servers = await createMultipleServers(2)
await setAccessTokensToServers(servers)
await setDefaultVideoChannel(servers)
await doubleFollow(servers[0], servers[1])
await servers[0].config.enableTranscription({ remote: true })
const registrationToken = await servers[0].runnerRegistrationTokens.getFirstRegistrationToken()
peertubeRunner = new PeerTubeRunnerProcess(servers[0])
await peertubeRunner.runServer()
await peertubeRunner.registerPeerTubeInstance({ registrationToken, runnerName: 'runner' })
})
describe('Running transcription', function () {
it('Should run transcription on classic file', async function () {
this.timeout(360000)
const uuid = await uploadForTranscription(servers[0])
await waitJobs(servers, { runnerJobs: true })
await checkCaption(servers, uuid)
await checkLanguage(servers, uuid, 'en')
})
it('Should not run transcription on video without audio stream', async function () {
this.timeout(360000)
const uuid = await uploadForTranscription(servers[0], { fixture: 'video_short_no_audio.mp4' })
await waitJobs(servers)
let continueWhile = true
while (continueWhile) {
await wait(500)
const { data } = await servers[0].runnerJobs.list({ stateOneOf: [ RunnerJobState.ERRORED ] })
continueWhile = !data.some(j => j.type === 'video-transcription')
}
await checkNoCaption(servers, uuid)
await checkLanguage(servers, uuid, null)
})
})
describe('Check cleanup', function () {
it('Should have an empty cache directory', async function () {
await checkPeerTubeRunnerCacheIsEmpty(peertubeRunner, 'transcription')
})
})
after(async function () {
if (peertubeRunner) {
await peertubeRunner.unregisterPeerTubeInstance({ runnerName: 'runner' })
peertubeRunner.kill()
}
await cleanupTests(servers)
})
})

View File

@ -373,7 +373,7 @@ describe('Test VOD transcoding in peertube-runner program', function () {
describe('Check cleanup', function () {
it('Should have an empty cache directory', async function () {
await checkPeerTubeRunnerCacheIsEmpty(peertubeRunner)
await checkPeerTubeRunnerCacheIsEmpty(peertubeRunner, 'transcoding')
})
})

View File

@ -32,13 +32,13 @@ export async function checkDirectoryIsEmpty (server: PeerTubeServer, directory:
expect(filtered).to.have.lengthOf(0)
}
export async function checkPeerTubeRunnerCacheIsEmpty (runner: PeerTubeRunnerProcess) {
const directoryPath = join(homedir(), '.cache', 'peertube-runner-nodejs', runner.getId(), 'transcoding')
export async function checkPeerTubeRunnerCacheIsEmpty (runner: PeerTubeRunnerProcess, subDir: 'transcoding' | 'transcription') {
const directoryPath = join(homedir(), '.cache', 'peertube-runner-nodejs', runner.getId(), subDir)
const directoryExists = await pathExists(directoryPath)
expect(directoryExists).to.be.true
const files = await readdir(directoryPath)
expect(files, 'Directory content: ' + files.join(', ')).to.have.lengthOf(0)
expect(files, `Sub-directory ${subDir} content: ${files.join(', ')}`).to.have.lengthOf(0)
}

View File

@ -26,6 +26,7 @@ export const FIXTURE_URLS = {
badVideo: 'https://download.cpy.re/peertube/bad_video.mp4',
goodVideo: 'https://download.cpy.re/peertube/good_video.mp4',
goodVideo720: 'https://download.cpy.re/peertube/good_video_720.mp4',
transcriptionVideo: 'https://download.cpy.re/peertube/the_last_man_on_earth.mp4',
chatersVideo: 'https://download.cpy.re/peertube/video_chapters.mp4',

View File

@ -54,6 +54,7 @@ function getAllNotificationsSettings (): UserNotificationSetting {
autoInstanceFollowing: UserNotificationSettingValue.WEB | UserNotificationSettingValue.EMAIL,
newPeerTubeVersion: UserNotificationSettingValue.WEB | UserNotificationSettingValue.EMAIL,
myVideoStudioEditionFinished: UserNotificationSettingValue.WEB | UserNotificationSettingValue.EMAIL,
myVideoTranscriptionGenerated: UserNotificationSettingValue.WEB | UserNotificationSettingValue.EMAIL,
newPluginVersion: UserNotificationSettingValue.WEB | UserNotificationSettingValue.EMAIL
}
}
@ -758,6 +759,40 @@ async function checkNewPluginVersion (options: CheckerBaseParams & {
await checkNotification({ ...options, notificationChecker, emailNotificationFinder })
}
async function checkMyVideoTranscriptionGenerated (options: CheckerBaseParams & {
videoName: string
shortUUID: string
language: {
id: string
label: string
}
checkType: CheckerType
}) {
const { videoName, shortUUID, language } = options
const notificationType = UserNotificationType.MY_VIDEO_TRANSCRIPTION_GENERATED
function notificationChecker (notification: UserNotification, checkType: CheckerType) {
if (checkType === 'presence') {
expect(notification).to.not.be.undefined
expect(notification.type).to.equal(notificationType)
expect(notification.videoCaption).to.exist
expect(notification.videoCaption.language.id).to.equal(language.id)
expect(notification.videoCaption.language.label).to.equal(language.label)
checkVideo(notification.videoCaption.video, videoName, shortUUID)
} else {
expect(notification.videoCaption).to.satisfy(c => c === undefined || c.Video.shortUUID !== shortUUID)
}
}
function emailNotificationFinder (email: object) {
const text: string = email['text']
return text.includes(shortUUID) && text.includes('Transcription in ' + language.label)
}
await checkNotification({ ...options, notificationChecker, emailNotificationFinder })
}
async function prepareNotificationsTest (serversCount = 3, overrideConfigArg: any = {}) {
const userNotifications: UserNotification[] = []
const adminNotifications: UserNotification[] = []
@ -863,7 +898,8 @@ export {
checkNewPeerTubeVersion,
checkNewPluginVersion,
checkVideoStudioEditionIsFinished,
checkRegistrationRequest
checkRegistrationRequest,
checkMyVideoTranscriptionGenerated
}
// ---------------------------------------------------------------------------

View File

@ -0,0 +1,81 @@
/* eslint-disable @typescript-eslint/no-unused-expressions,@typescript-eslint/require-await */
import { buildAbsoluteFixturePath } from '@peertube/peertube-node-utils'
import { PeerTubeServer, VideoEdit } from '@peertube/peertube-server-commands'
import { downloadFile, unzip } from '@peertube/peertube-transcription-devtools'
import { expect } from 'chai'
import { ensureDir, pathExists } from 'fs-extra/esm'
import { join } from 'path'
import { testCaptionFile } from './captions.js'
import { FIXTURE_URLS } from './fixture-urls.js'
type CustomModelName = 'tiny.pt' | 'faster-whisper-tiny'
export async function downloadCustomModelsIfNeeded (modelName: CustomModelName) {
if (await pathExists(getCustomModelPath(modelName))) return
await ensureDir(getCustomModelDirectory())
await unzip(await downloadFile(FIXTURE_URLS.transcriptionModels, getCustomModelDirectory()))
}
export function getCustomModelDirectory () {
return buildAbsoluteFixturePath(join('transcription', 'models-v1'))
}
export function getCustomModelPath (modelName: CustomModelName) {
return join(getCustomModelDirectory(), 'models', modelName)
}
// ---------------------------------------------------------------------------
export async function checkCaption (servers: PeerTubeServer[], uuid: string, captionContains = 'WEBVTT\n\n00:00.000 --> 00:') {
for (const server of servers) {
const body = await server.captions.list({ videoId: uuid })
expect(body.total).to.equal(1)
expect(body.data).to.have.lengthOf(1)
const caption = body.data[0]
expect(caption.language.id).to.equal('en')
expect(caption.language.label).to.equal('English')
{
await testCaptionFile(server.url, caption.captionPath, captionContains)
}
}
}
export async function checkNoCaption (servers: PeerTubeServer[], uuid: string) {
for (const server of servers) {
const body = await server.captions.list({ videoId: uuid })
expect(body.total).to.equal(0)
expect(body.data).to.have.lengthOf(0)
}
}
// ---------------------------------------------------------------------------
export async function checkLanguage (servers: PeerTubeServer[], uuid: string, expected: string | null) {
for (const server of servers) {
const video = await server.videos.get({ id: uuid })
if (expected) {
expect(video.language.id).to.equal(expected)
} else {
expect(video.language.id).to.be.null
}
}
}
export async function uploadForTranscription (server: PeerTubeServer, body: Partial<VideoEdit> = {}) {
const { uuid } = await server.videos.upload({
attributes: {
name: 'video',
fixture: join('transcription', 'videos', 'the_last_man_on_earth.mp4'),
language: undefined,
...body
}
})
return uuid
}

View File

@ -1,5 +1,5 @@
import { levenshteinDistance } from '@peertube/peertube-transcription-devtools'
import { expect } from 'chai'
import { levenshteinDistance } from '@peertube/peertube-transcription'
describe('Levenshtein distance', function () {
it(`equals 1 when there is only one character difference`, function () {

View File

@ -1,17 +1,16 @@
import { transcriberFactory } from '@peertube/peertube-transcription'
import { TranscriptionEngineName, transcriberFactory } from '@peertube/peertube-transcription'
import { createLogger } from 'winston'
describe('Transcriber factory', function () {
const transcribers = [
'openai-whisper',
'whisper-ctranslate2',
'whisper-timestamped'
]
const transcribers: TranscriptionEngineName[] = [ 'openai-whisper', 'whisper-ctranslate2' ]
describe('Should be able to create a transcriber for each available transcription engine', function () {
transcribers.forEach(function (transcriberName) {
for (const transcriberName of transcribers) {
it(`Should be able to create a(n) ${transcriberName} transcriber`, function () {
transcriberFactory.createFromEngineName(transcriberName)
})
transcriberFactory.createFromEngineName({ engineName: transcriberName, logger: createLogger() })
})
}
})
})

View File

@ -1,17 +1,18 @@
/* eslint-disable @typescript-eslint/no-unused-expressions, no-new, max-len */
import { TranscriptFile, TranscriptFileEvaluator } from '@peertube/peertube-transcription'
import { buildAbsoluteFixturePath } from '@peertube/peertube-node-utils'
import { join } from 'node:path'
import { mkdir, rm } from 'node:fs/promises'
import { tmpdir } from 'node:os'
import { TranscriptFile } from '@peertube/peertube-transcription'
import { TranscriptFileEvaluator } from '@peertube/peertube-transcription-devtools'
import { expect } from 'chai'
import { ensureDir, remove } from 'fs-extra/esm'
import { tmpdir } from 'node:os'
import { join } from 'node:path'
describe('Transcript File Evaluator', function () {
const transcriptDirectory = join(tmpdir(), 'peertube-transcription', 'transcript-file-evaluator')
const referenceTranscriptFilePath = buildAbsoluteFixturePath('transcription/videos/communiquer-lors-dune-classe-transplantee.txt')
before(async function () {
await mkdir(transcriptDirectory, { recursive: true })
await ensureDir(transcriptDirectory)
})
it(`may not compare files in another format than txt`, async function () {
@ -62,6 +63,6 @@ Ensuite, il pourront lire et commenter ce de leurs camarades ou répondre aux co
})
after(async function () {
await rm(transcriptDirectory, { recursive: true, force: true })
await remove(transcriptDirectory)
})
})

View File

@ -1,15 +1,16 @@
/* eslint-disable @typescript-eslint/no-unused-expressions */
import { expect } from 'chai'
import { join } from 'node:path'
import { mkdir, rm } from 'node:fs/promises'
import { TranscriptFile } from '@peertube/peertube-transcription'
import { tmpdir } from 'node:os'
import { buildAbsoluteFixturePath } from '@peertube/peertube-node-utils'
import { TranscriptFile } from '@peertube/peertube-transcription'
import { expect } from 'chai'
import { ensureDir, remove } from 'fs-extra/esm'
import { tmpdir } from 'node:os'
import { join } from 'node:path'
describe('Transcript File', function () {
const transcriptFileDirectory = join(tmpdir(), 'peertube-transcription', 'transcript-file')
before(async function () {
await mkdir(transcriptFileDirectory, { recursive: true })
await ensureDir(transcriptFileDirectory)
})
it(`may creates a new transcript file from scratch`, async function () {
@ -33,12 +34,11 @@ describe('Transcript File', function () {
})
it(`fails when loading a file which is obviously not a transcript`, function () {
expect(() => TranscriptFile.fromPath(buildAbsoluteFixturePath('transcription/videos/the_last_man_on_earth.mp4'), 'en'))
.to.throw(`Couldn't guess transcript format from extension "mp4". Valid formats are: txt, vtt, srt.`)
})
after(async function () {
await rm(transcriptFileDirectory, { recursive: true, force: true })
await remove(transcriptFileDirectory)
})
})

View File

@ -1 +0,0 @@
describe('Transcription run', function () {})

View File

@ -1,14 +1,16 @@
import { cp, lstat, mkdir, rm } from 'node:fs/promises'
import { join } from 'node:path'
import { tmpdir } from 'node:os'
import { expect } from 'chai'
import { downloadFile, unzip } from '@peertube/peertube-transcription'
import { buildAbsoluteFixturePath } from '@peertube/peertube-node-utils'
import { downloadFile, unzip } from '@peertube/peertube-transcription-devtools'
import { expect } from 'chai'
import { ensureDir, remove } from 'fs-extra/esm'
import { cp, lstat } from 'node:fs/promises'
import { tmpdir } from 'node:os'
import { join } from 'node:path'
describe('downloadFile', function () {
const testDirectory = join(tmpdir(), 'peertube-transcription', 'utils')
before(async function () {
await mkdir(testDirectory, { recursive: true })
await ensureDir(testDirectory)
})
it(`Downloads a file and write it to the disk `, async function () {
@ -18,7 +20,7 @@ describe('downloadFile', function () {
})
after(async function () {
await rm(testDirectory, { recursive: true, force: true })
await remove(testDirectory)
})
})
@ -26,8 +28,9 @@ describe('unzip', function () {
const zipFixtureFileName = 'hello_world.zip'
const zipFixtureFilePath = buildAbsoluteFixturePath(`transcription/${zipFixtureFileName}`)
const testDirectory = join(tmpdir(), 'peertube-transcription', 'utils')
before(async function () {
await mkdir(testDirectory, { recursive: true })
await ensureDir(testDirectory)
})
it(`Extract zip archive to directory`, async function () {
@ -39,6 +42,6 @@ describe('unzip', function () {
})
after(async function () {
await rm(testDirectory, { recursive: true, force: true })
await remove(testDirectory)
})
})

View File

@ -1,57 +1,59 @@
/* eslint-disable @typescript-eslint/no-unused-expressions, max-len */
import { expect, config } from 'chai'
import { createLogger } from 'winston'
import { join } from 'node:path'
import { mkdir, rm } from 'node:fs/promises'
import { tmpdir } from 'node:os'
import { buildAbsoluteFixturePath } from '@peertube/peertube-node-utils'
import {
downloadFile,
levenshteinDistance,
OpenaiTranscriber,
TranscriptFile,
TranscriptFileEvaluator,
TranscriptionModel,
unzip,
WhisperBuiltinModel
} from '@peertube/peertube-transcription'
import { FIXTURE_URLS } from '@tests/shared/fixture-urls.js'
import { TranscriptFileEvaluator, levenshteinDistance } from '@peertube/peertube-transcription-devtools'
import { downloadCustomModelsIfNeeded, getCustomModelPath } from '@tests/shared/transcription.js'
import { config, expect } from 'chai'
import { ensureDir, remove } from 'fs-extra/esm'
import { tmpdir } from 'node:os'
import { join } from 'node:path'
import { createLogger } from 'winston'
config.truncateThreshold = 0
describe('Open AI Whisper transcriber', function () {
const tmpDirectory = join(tmpdir(), 'peertube-transcription')
const transcriptDirectory = join(tmpDirectory, 'transcriber', 'openai')
const modelsDirectory = join(tmpDirectory, 'models')
const shortVideoPath = buildAbsoluteFixturePath('transcription/videos/the_last_man_on_earth.mp4')
const frVideoPath = buildAbsoluteFixturePath('transcription/videos/derive_sectaire.mp4')
const referenceTranscriptFile = new TranscriptFile({
path: buildAbsoluteFixturePath('transcription/videos/derive_sectaire.txt'),
language: 'fr',
format: 'txt'
})
const transcriber = new OpenaiTranscriber(
{
const transcriber = new OpenaiTranscriber({
engine: {
name: 'openai-whisper',
requirements: [],
type: 'binary',
binary: 'whisper',
command: 'whisper',
supportedModelFormats: [ 'PyTorch' ],
languageDetection: true
languageDetection: true,
version: ''
},
createLogger(),
transcriptDirectory
)
logger: createLogger()
})
const model = new TranscriptionModel('tiny')
before(async function () {
this.timeout(1 * 1000 * 60)
await mkdir(transcriptDirectory, { recursive: true })
await unzip(await downloadFile(FIXTURE_URLS.transcriptionModels, tmpDirectory))
this.timeout(120000)
await ensureDir(transcriptDirectory)
await downloadCustomModelsIfNeeded('tiny.pt')
})
it('Should transcribe a media file and provide a valid path to a transcript file in `vtt` format by default', async function () {
it('Should transcribe a media file and provide a valid path to a transcript file in `vtt` format', async function () {
this.timeout(3 * 1000 * 60)
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en' })
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'vtt', model, transcriptDirectory })
expect(transcript.format).to.equals('vtt')
expect(transcript.language).to.equals('en')
@ -59,7 +61,7 @@ describe('Open AI Whisper transcriber', function () {
})
it('May produce a transcript file in the `srt` format', async function () {
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'srt' })
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'srt', model, transcriptDirectory })
expect(transcript.format).to.equals('srt')
expect(transcript.language).to.equals('en')
@ -67,7 +69,7 @@ describe('Open AI Whisper transcriber', function () {
})
it('May produce a transcript file in the `txt` format', async function () {
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'txt' })
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'txt', model, transcriptDirectory })
expect(transcript.format).to.equals('txt')
expect(transcript.language).to.equals('en')
@ -80,16 +82,20 @@ describe('Open AI Whisper transcriber', function () {
it('May transcribe a media file using a local PyTorch model', async function () {
this.timeout(2 * 1000 * 60)
await transcriber.transcribe({
mediaFilePath: shortVideoPath,
model: await TranscriptionModel.fromPath(join(modelsDirectory, 'tiny.pt')),
language: 'en'
model: await TranscriptionModel.fromPath(getCustomModelPath('tiny.pt')),
language: 'en',
format: 'vtt',
transcriptDirectory
})
})
it('May transcribe a media file in french', async function () {
this.timeout(3 * 1000 * 60)
const transcript = await transcriber.transcribe({ mediaFilePath: frVideoPath, language: 'fr', format: 'txt' })
const transcript = await transcriber.transcribe({ mediaFilePath: frVideoPath, language: 'fr', format: 'txt', model, transcriptDirectory })
expect(transcript.format).to.equals('txt')
expect(transcript.language).to.equals('fr')
@ -98,18 +104,21 @@ describe('Open AI Whisper transcriber', function () {
it('Guesses the video language if not provided', async function () {
this.timeout(3 * 1000 * 60)
const transcript = await transcriber.transcribe({ mediaFilePath: frVideoPath })
const transcript = await transcriber.transcribe({ mediaFilePath: frVideoPath, model, format: 'vtt', transcriptDirectory })
expect(transcript.language).to.equals('fr')
})
it('May transcribe a media file in french with small model', async function () {
it('May transcribe a media file in french with small model (can be long)', async function () {
this.timeout(6 * 1000 * 60)
const transcript = await transcriber.transcribe({
mediaFilePath: frVideoPath,
language: 'fr',
format: 'txt',
model: new WhisperBuiltinModel('small')
model: new WhisperBuiltinModel('small'),
transcriptDirectory
})
expect(transcript.language).to.equals('fr')
@ -120,6 +129,6 @@ describe('Open AI Whisper transcriber', function () {
})
after(async function () {
await rm(transcriptDirectory, { recursive: true, force: true })
await remove(transcriptDirectory)
})
})

View File

@ -1,133 +0,0 @@
/* eslint-disable @typescript-eslint/no-unused-expressions, max-len */
import { expect, config } from 'chai'
import { createLogger } from 'winston'
import { join } from 'node:path'
import { mkdir, rm } from 'node:fs/promises'
import { tmpdir } from 'node:os'
import { buildAbsoluteFixturePath } from '@peertube/peertube-node-utils'
import {
OpenaiTranscriber,
WhisperTimestampedTranscriber,
TranscriptFileEvaluator,
TranscriptionModel,
WhisperTranscribeArgs,
levenshteinDistance, downloadFile, unzip
} from '@peertube/peertube-transcription'
import { FIXTURE_URLS } from '@tests/shared/fixture-urls.js'
config.truncateThreshold = 0
describe('Linto timestamped Whisper transcriber', function () {
const tmpDirectory = join(tmpdir(), 'peertube-transcription')
const transcriptDirectory = join(tmpDirectory, 'transcriber', 'timestamped')
const modelsDirectory = join(tmpDirectory, 'models')
const shortVideoPath = buildAbsoluteFixturePath('transcription/videos/the_last_man_on_earth.mp4')
const frVideoPath = buildAbsoluteFixturePath('transcription/videos/derive_sectaire.mp4')
const transcriber = new WhisperTimestampedTranscriber(
{
name: 'whisper-timestamped',
requirements: [],
type: 'binary',
binary: 'whisper_timestamped',
supportedModelFormats: [ 'PyTorch' ],
languageDetection: true
},
createLogger(),
transcriptDirectory
)
before(async function () {
this.timeout(1 * 1000 * 60)
await mkdir(transcriptDirectory, { recursive: true })
await unzip(await downloadFile(FIXTURE_URLS.transcriptionModels, tmpDirectory))
})
it('Should transcribe a media file and provide a valid path to a transcript file in `vtt` format by default', async function () {
this.timeout(1 * 1000 * 60)
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en' })
expect(transcript.format).to.equals('vtt')
expect(transcript.language).to.equals('en')
expect(await transcript.read()).not.to.be.empty
})
it('May produce a transcript file in the `srt` format with a ms precision', async function () {
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'srt' })
expect(transcript.format).to.equals('srt')
expect(transcript.language).to.equals('en')
expect(await transcript.read()).not.to.be.empty
})
it('May produce a transcript file in `txt` format', async function () {
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'txt' })
expect(transcript.format).to.equals('txt')
expect(transcript.language).to.equals('en')
expect(await transcript.read()).not.to.be.empty
expect(levenshteinDistance(
(await transcript.read()).toString(),
'December 1965, is that all it has been since I inherited the world only three years, seems like a hundred million.'
)).to.be.below(10)
})
it('May transcribe a media file using a local PyTorch model file', async function () {
this.timeout(2 * 1000 * 60)
await transcriber.transcribe({
mediaFilePath: shortVideoPath,
model: await TranscriptionModel.fromPath(join(modelsDirectory, 'tiny.pt')),
language: 'en'
})
})
it('May transcribe a media file in french', async function () {
this.timeout(2 * 1000 * 60)
const transcript = await transcriber.transcribe({
mediaFilePath: frVideoPath,
language: 'fr',
format: 'txt'
})
expect(transcript.format).to.equals('txt')
expect(transcript.language).to.equals('fr')
expect(await transcript.read()).not.to.be.empty
})
it('Guesses the video language if not provided', async function () {
this.timeout(2 * 1000 * 60)
const transcript = await transcriber.transcribe({ mediaFilePath: frVideoPath })
expect(transcript.language).to.equals('fr')
})
it('Should produce a text transcript similar to openai-whisper implementation', async function () {
this.timeout(11 * 1000 * 60)
const transcribeArgs: WhisperTranscribeArgs = {
mediaFilePath: frVideoPath,
model: await TranscriptionModel.fromPath(join(modelsDirectory, 'tiny.pt')),
language: 'fr',
format: 'txt'
}
const transcript = await transcriber.transcribe(transcribeArgs)
const openaiTranscriber = new OpenaiTranscriber(
{
name: 'openai-whisper',
requirements: [],
type: 'binary',
binary: 'whisper',
supportedModelFormats: [ 'PyTorch' ]
},
createLogger(),
join(transcriptDirectory, 'openai-whisper')
)
const openaiTranscript = await openaiTranscriber.transcribe(transcribeArgs)
const transcriptFileEvaluator = new TranscriptFileEvaluator(openaiTranscript, transcript)
expect(await transcriptFileEvaluator.wer()).to.be.below(25 / 100)
expect(await transcriptFileEvaluator.cer()).to.be.below(15 / 100)
})
after(async function () {
await rm(transcriptDirectory, { recursive: true, force: true })
})
})

View File

@ -1,50 +1,58 @@
/* eslint-disable @typescript-eslint/no-unused-expressions, max-len */
import { expect, config } from 'chai'
import { createLogger } from 'winston'
import { join } from 'node:path'
import { mkdir, rm } from 'node:fs/promises'
import { tmpdir } from 'node:os'
import { buildAbsoluteFixturePath } from '@peertube/peertube-node-utils'
import {
Ctranslate2Transcriber, downloadFile,
levenshteinDistance,
Ctranslate2Transcriber,
OpenaiTranscriber,
TranscriptFile,
TranscriptFileEvaluator,
TranscriptionModel, unzip,
WhisperTranscribeArgs
TranscriptionModel
} from '@peertube/peertube-transcription'
import { FIXTURE_URLS } from '@tests/shared/fixture-urls.js'
import { TranscriptFileEvaluator, levenshteinDistance } from '@peertube/peertube-transcription-devtools'
import { downloadCustomModelsIfNeeded, getCustomModelPath } from '@tests/shared/transcription.js'
import { config, expect } from 'chai'
import { ensureDir, remove } from 'fs-extra/esm'
import { tmpdir } from 'node:os'
import { join } from 'node:path'
import { createLogger } from 'winston'
config.truncateThreshold = 0
describe('Whisper CTranslate2 transcriber', function () {
const tmpDirectory = join(tmpdir(), 'peertube-transcription')
const transcriptDirectory = join(tmpDirectory, 'transcriber', 'ctranslate2')
const modelsDirectory = join(tmpDirectory, 'models')
const shortVideoPath = buildAbsoluteFixturePath('transcription/videos/the_last_man_on_earth.mp4')
const frVideoPath = buildAbsoluteFixturePath('transcription/videos/derive_sectaire.mp4')
const transcriber = new Ctranslate2Transcriber(
{
name: 'anyNameShouldBeFineReally',
requirements: [],
type: 'binary',
binary: 'whisper-ctranslate2',
supportedModelFormats: [],
languageDetection: true
},
createLogger(),
transcriptDirectory
)
before(async function () {
this.timeout(1 * 1000 * 60)
await mkdir(transcriptDirectory, { recursive: true })
await unzip(await downloadFile(FIXTURE_URLS.transcriptionModels, tmpDirectory))
const transcriber = new Ctranslate2Transcriber({
engine: {
name: 'whisper-ctranslate2',
type: 'binary',
command: 'whisper-ctranslate2',
supportedModelFormats: [ 'CTranslate2' ],
languageDetection: true,
version: '0.4.4'
},
logger: createLogger()
})
it('Should transcribe a media file and provide a valid path to a transcript file in `vtt` format by default', async function () {
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en' })
const model = new TranscriptionModel('tiny')
before(async function () {
this.timeout(120000)
await ensureDir(transcriptDirectory)
await downloadCustomModelsIfNeeded('faster-whisper-tiny')
})
it('Should transcribe a media file and provide a valid path to a transcript file in `vtt` format', async function () {
const transcript = await transcriber.transcribe({
mediaFilePath: shortVideoPath,
language: 'en',
format: 'vtt',
model,
transcriptDirectory
})
expect(transcript.format).to.equals('vtt')
expect(transcript.language).to.equals('en')
@ -52,7 +60,13 @@ describe('Whisper CTranslate2 transcriber', function () {
})
it('May produce a transcript file in the `srt` format', async function () {
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'srt' })
const transcript = await transcriber.transcribe({
mediaFilePath: shortVideoPath,
language: 'en',
format: 'srt',
model,
transcriptDirectory
})
expect(transcript.format).to.equals('srt')
expect(transcript.language).to.equals('en')
@ -60,7 +74,13 @@ describe('Whisper CTranslate2 transcriber', function () {
})
it('May produce a transcript file in the `txt` format', async function () {
const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'txt' })
const transcript = await transcriber.transcribe({
mediaFilePath: shortVideoPath,
language: 'en',
format: 'txt',
model,
transcriptDirectory
})
expect(await transcript.equals(new TranscriptFile({
path: join(transcriptDirectory, 'the_last_man_on_earth.txt'),
format: 'txt',
@ -78,10 +98,12 @@ describe('Whisper CTranslate2 transcriber', function () {
it('May transcribe a media file using a local CTranslate2 model', async function () {
this.timeout(2 * 1000 * 60)
const transcript = await transcriber.transcribe({
mediaFilePath: shortVideoPath,
model: await TranscriptionModel.fromPath(join(modelsDirectory, 'faster-whisper-tiny')),
model: await TranscriptionModel.fromPath(getCustomModelPath('faster-whisper-tiny')),
language: 'en',
transcriptDirectory,
format: 'txt'
})
@ -92,7 +114,14 @@ describe('Whisper CTranslate2 transcriber', function () {
it('May transcribe a media file in french', async function () {
this.timeout(5 * 1000 * 60)
const transcript = await transcriber.transcribe({ mediaFilePath: frVideoPath, language: 'fr', format: 'txt' })
const transcript = await transcriber.transcribe({
mediaFilePath: frVideoPath,
language: 'fr',
format: 'txt',
model,
transcriptDirectory
})
expect(transcript.format).to.equals('txt')
expect(transcript.language).to.equals('fr')
@ -101,30 +130,39 @@ describe('Whisper CTranslate2 transcriber', function () {
it('Guesses the video language if not provided', async function () {
this.timeout(2 * 1000 * 60)
const transcript = await transcriber.transcribe({ mediaFilePath: frVideoPath })
const transcript = await transcriber.transcribe({ mediaFilePath: frVideoPath, format: 'vtt', model, transcriptDirectory })
expect(transcript.language).to.equals('fr')
})
it('Should produce a text transcript similar to openai-whisper implementation', async function () {
this.timeout(10 * 1000 * 60)
const transcribeArgs: WhisperTranscribeArgs = {
const transcribeArgs = {
mediaFilePath: frVideoPath,
language: 'fr',
format: 'txt'
format: 'txt' as 'txt',
transcriptDirectory,
model
}
const transcript = await transcriber.transcribe(transcribeArgs)
const openaiTranscriber = new OpenaiTranscriber(
{
const openaiTranscriber = new OpenaiTranscriber({
engine: {
name: 'openai-whisper',
requirements: [],
type: 'binary',
binary: 'whisper',
supportedModelFormats: [ 'PyTorch' ]
command: 'whisper',
supportedModelFormats: [ 'PyTorch' ],
version: '0.4.4'
},
createLogger(),
join(transcriptDirectory, 'openai-whisper')
)
const openaiTranscript = await openaiTranscriber.transcribe(transcribeArgs)
logger: createLogger()
})
const openaiTranscript = await openaiTranscriber.transcribe({
...transcribeArgs,
transcriptDirectory: join(transcriptDirectory, 'openai-whisper')
})
const transcriptFileEvaluator = new TranscriptFileEvaluator(openaiTranscript, transcript)
expect(await transcriptFileEvaluator.wer()).to.be.below(20 / 100)
@ -132,6 +170,6 @@ describe('Whisper CTranslate2 transcriber', function () {
})
after(async function () {
await rm(transcriptDirectory, { recursive: true, force: true })
await remove(transcriptDirectory)
})
})

View File

@ -8,13 +8,13 @@
"@tests/*": [ "./src/*" ],
"@server/*": [ "../../server/core/*" ],
"@peertube/peertube-transcription": [ "../transcription" ],
"@peertube/peertube-jiwer": [ "../jiwer" ],
"@peertube/peertube-transcription-devtools": [ "../transcription-devtools" ],
}
},
"references": [
{ "path": "../core-utils" },
{ "path": "../ffmpeg" },
{ "path": "../jiwer" },
{ "path": "../transcription-devtools" },
{ "path": "../models" },
{ "path": "../node-utils" },
{ "path": "../typescript-utils" },

View File

@ -0,0 +1,63 @@
# Transcription DevTools
Includes:
* __JiWER__ CLI NodeJS wrapper
* Benchmark tool to test multiple transcription engines
* TypeScript classes to evaluate word-error-rate of files generated by the transcription
## Build
```sh
npm run build
```
## Benchmark
A benchmark of available __transcribers__ might be run with:
```sh
npm run benchmark
```
```
┌────────────────────────┬───────────────────────┬───────────────────────┬──────────┬────────┬───────────────────────┐
│ (index) │ WER │ CER │ duration │ model │ engine │
├────────────────────────┼───────────────────────┼───────────────────────┼──────────┼────────┼───────────────────────┤
│ 5yZGBYqojXe7nuhq1TuHvz │ '28.39506172839506%' │ '9.62457337883959%' │ '41s' │ 'tiny' │ 'openai-whisper' │
│ x6qREJ2AkTU4e5YmvfivQN │ '29.75206611570248%' │ '10.46195652173913%' │ '15s' │ 'tiny' │ 'whisper-ctranslate2' │
└────────────────────────┴───────────────────────┴───────────────────────┴──────────┴────────┴───────────────────────┘
```
The benchmark may be run with multiple model builtin sizes:
```sh
MODELS=tiny,small,large npm run benchmark
```
## Jiwer
> *JiWER is a python tool for computing the word-error-rate of ASR systems.*
> https://jitsi.github.io/jiwer/cli/
__JiWER__ serves as a reference implementation to calculate errors rates between 2 text files:
- WER (Word Error Rate)
- CER (Character Error Rate)
### Usage
```typescript
const jiwerCLI = new JiwerClI('./reference.txt', './hypothesis.txt')
// WER as a percentage, ex: 0.03 -> 3%
console.log(await jiwerCLI.wer())
// CER as a percentage: 0.01 -> 1%
console.log(await jiwerCLI.cer())
// Detailed comparison report
console.log(await jiwerCLI.alignment())
```
## Resources
- https://jitsi.github.io/jiwer/
- https://github.com/rapidfuzz/RapidFuzz

View File

@ -1,5 +1,5 @@
{
"name": "@peertube/peertube-jiwer",
"name": "@peertube/peertube-transcription-devtools",
"private": true,
"version": "0.0.0",
"main": "dist/index.js",
@ -12,9 +12,9 @@
"type": "module",
"devDependencies": {},
"scripts": {
"preinstall": "pip install -r requirements.txt",
"build": "tsc",
"watch": "tsc -w"
"watch": "tsc -w",
"benchmark": "tsx --conditions=peertube:tsx --tsconfig ./tsconfig.json ./src/benchmark.ts"
},
"dependencies": {}
}

View File

@ -0,0 +1 @@
jiwer

View File

@ -1,17 +1,18 @@
import { createLogger, transports, format } from 'winston'
import { join } from 'node:path'
import { performance, PerformanceObserver } from 'node:perf_hooks'
import { tmpdir } from 'node:os'
import { rm, mkdir } from 'node:fs/promises'
import { buildAbsoluteFixturePath, buildSUUID, SUUID } from '@peertube/peertube-node-utils'
import {
transcriberFactory,
TranscriptFile,
TranscriptFileEvaluator,
TranscriptionEngine,
TranscriptionModel
} from '@peertube/peertube-transcription'
import { millisecondsToTime } from '@peertube/peertube-core-utils'
import { SUUID, buildAbsoluteFixturePath, buildSUUID } from '@peertube/peertube-node-utils'
import {
TranscriptFile,
TranscriptionEngine,
TranscriptionEngineName,
TranscriptionModel,
transcriberFactory
} from '@peertube/peertube-transcription'
import { ensureDir, remove } from 'fs-extra/esm'
import { tmpdir } from 'node:os'
import { join } from 'node:path'
import { PerformanceObserver, performance } from 'node:perf_hooks'
import { createLogger, format, transports } from 'winston'
import { TranscriptFileEvaluator } from './transcript-file-evaluator.js'
interface BenchmarkResult {
uuid: SUUID
@ -59,16 +60,15 @@ const formatBenchmarkResult = ({ WER, CER, duration, engine, model }: Partial<Be
void (async () => {
const logger = createLogger()
logger.add(new transports.Console({ format: format.printf(log => log.message) }))
const transcribers = [
'openai-whisper',
'whisper-ctranslate2',
'whisper-timestamped'
]
const transcribers: TranscriptionEngineName[] = [ 'openai-whisper', 'whisper-ctranslate2' ]
const models = process.env.MODELS
? process.env.MODELS.trim().split(',').map(modelName => modelName.trim()).filter(modelName => modelName)
: [ 'tiny' ]
const transcriptDirectory = join(tmpdir(), 'peertube-transcription', 'benchmark')
const pipDirectory = join(tmpdir(), 'peertube-transcription', 'pip')
const mediaFilePath = buildAbsoluteFixturePath('transcription/videos/derive_sectaire.mp4')
const referenceTranscriptFile = new TranscriptFile({
path: buildAbsoluteFixturePath('transcription/videos/derive_sectaire.txt'),
@ -79,7 +79,7 @@ void (async () => {
let benchmarkResults: Record<string, BenchmarkResult> = {}
// before
await mkdir(transcriptDirectory, { recursive: true })
await ensureDir(transcriptDirectory)
const performanceObserver = new PerformanceObserver((items) => {
items
.getEntries()
@ -97,11 +97,13 @@ void (async () => {
for (const transcriberName of transcribers) {
logger.info(`Create "${transcriberName}" transcriber for the benchmark...`)
const transcriber = transcriberFactory.createFromEngineName(
transcriberName,
createLogger(),
transcriptDirectory
)
const transcriber = transcriberFactory.createFromEngineName({
engineName: transcriberName,
logger: createLogger(),
binDirectory: join(pipDirectory, 'bin')
})
await transcriber.install(pipDirectory)
for (const modelName of models) {
logger.info(`Run benchmark with "${modelName}" model:`)
@ -110,6 +112,7 @@ void (async () => {
const transcriptFile = await transcriber.transcribe({
mediaFilePath,
model,
transcriptDirectory,
language: 'fr',
format: 'txt',
runId: uuid
@ -134,6 +137,6 @@ void (async () => {
Object.values(benchmarkResultsGroupedByModel).forEach(benchmark => console.table(benchmark))
// after
await rm(transcriptDirectory, { recursive: true, force: true })
await remove(transcriptDirectory)
performance.clearMarks()
})()

View File

@ -0,0 +1,5 @@
export * from './jiwer-cli.js'
export * from './levenshtein.js'
export * from './transcript-file-evaluator-interface.js'
export * from './transcript-file-evaluator.js'
export * from './utils.js'

Some files were not shown because too many files have changed in this diff Show More