chore(transcription): change transcribers transcribe method signature

Introduce whisper builtin model.

packages/tests/src/transcription/benchmark.spec.ts

@@ -9,7 +9,8 @@ import {
   transcriberFactory,
   TranscriptFile,
   TranscriptFileEvaluator,
-  TranscriptionEngine
+  TranscriptionEngine,
+  TranscriptionModel
 } from '@peertube/peertube-transcription'
 
 interface TestResult {
@@ -101,9 +102,15 @@ describe('Transcribers benchmark', function () {
       models.forEach((modelName) => {
         it(`Run ${transcriberName} transcriber benchmark with ${modelName} model`, async function () {
           this.timeout(15 * 1000 * 60) // 15 minutes
-          const model = { name: modelName }
+          const model = new TranscriptionModel(modelName)
           const uuid = short.generate()
-          const transcriptFile = await transcriber.transcribe(mediaFilePath, model, 'fr', 'txt', uuid)
+          const transcriptFile = await transcriber.transcribe({
+            mediaFilePath,
+            model,
+            language: 'fr',
+            format: 'txt',
+            runId: uuid
+          })
           const evaluator = new TranscriptFileEvaluator(referenceTranscriptFile, transcriptFile)
           await new Promise(resolve => setTimeout(resolve, 1))
 

packages/tests/src/transcription/transcript/transcript-evaluator.spec.ts

@@ -7,7 +7,7 @@ import { expect } from 'chai'
 
 describe('Transcript File Evaluator', function () {
   const transcriptDirectory = buildAbsoluteFixturePath('transcription/transcript-evaluator')
-  const referenceTranscriptFilepath = buildAbsoluteFixturePath('transcription/transcript/reference.txt')
+  const referenceTranscriptFilePath = buildAbsoluteFixturePath('transcription/transcript/reference.txt')
 
   before(async function () {
     await mkdir(transcriptDirectory, { recursive: true })
@@ -29,7 +29,7 @@ describe('Transcript File Evaluator', function () {
 
   it(`evaluation must return coherent wer & cer`, async function () {
     const reference = new TranscriptFile({
-      path: referenceTranscriptFilepath,
+      path: referenceTranscriptFilePath,
       language: 'fr',
       format: 'txt'
     })

packages/tests/src/transcription/whisper/transcriber/openai-transcriber.spec.ts

@@ -4,15 +4,14 @@ import { createLogger } from 'winston'
 import { join } from 'path'
 import { mkdir, rm } from 'node:fs/promises'
 import { buildAbsoluteFixturePath, root } from '@peertube/peertube-node-utils'
-import { OpenaiTranscriber, TranscriptFile } from '@peertube/peertube-transcription'
+import { OpenaiTranscriber, TranscriptFile, TranscriptionModel, WhisperBuiltinModel } from '@peertube/peertube-transcription'
 
 config.truncateThreshold = 0
 
 describe('Open AI Whisper transcriber', function () {
   const transcriptDirectory = join(root(), 'test-transcript')
   const shortVideoPath = buildAbsoluteFixturePath('video_short.mp4')
   const frVideoPath = buildAbsoluteFixturePath('transcription/videos/communiquer-lors-dune-classe-transplantee.mp4')
-
   const transcriber = new OpenaiTranscriber(
     {
       name: 'openai-whisper',
@@ -30,7 +29,7 @@ describe('Open AI Whisper transcriber', function () {
   })
 
   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(shortVideoPath)
+    const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en' })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'video_short.vtt'),
       language: 'en',
@@ -48,7 +47,7 @@ You
   })
 
   it('May produce a transcript file in the `srt` format', async function () {
-    const transcript = await transcriber.transcribe(shortVideoPath, { name: 'tiny' }, 'en', 'srt')
+    const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'srt' })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'video_short.srt'),
       language: 'en',
@@ -65,7 +64,7 @@ You
   })
 
   it('May produce a transcript file in the `txt` format', async function () {
-    const transcript = await transcriber.transcribe(shortVideoPath, { name: 'tiny' }, 'en', 'txt')
+    const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'txt' })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'video_short.txt'),
       language: 'en',
@@ -77,12 +76,13 @@ You
   })
 
   it('May transcribe a media file using a local PyTorch model', async function () {
-    await transcriber.transcribe(frVideoPath, { name: 'myLocalModel', path: buildAbsoluteFixturePath('transcription/models/tiny.pt') }, 'fr')
+    this.timeout(2 * 1000 * 60)
+    await transcriber.transcribe({ mediaFilePath: frVideoPath, model: TranscriptionModel.fromPath(buildAbsoluteFixturePath('transcription/models/tiny.pt')), language: 'en' })
   })
 
   it('May transcribe a media file in french', async function () {
-    this.timeout(45000)
-    const transcript = await transcriber.transcribe(frVideoPath, { name: 'tiny' }, 'fr', 'txt')
+    this.timeout(2 * 1000 * 60)
+    const transcript = await transcriber.transcribe({ mediaFilePath: frVideoPath, language: 'fr', format: 'txt' })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'communiquer-lors-dune-classe-transplantee.txt'),
       language: 'fr',
@@ -105,8 +105,8 @@ Ensuite, il pourront lire et commenter ce de leurs camarades ou répondre aux co
   })
 
   it('May transcribe a media file in french with small model', async function () {
-    this.timeout(400000)
-    const transcript = await transcriber.transcribe(frVideoPath, { name: 'small' }, 'fr', 'txt')
+    this.timeout(5 * 1000 * 60)
+    const transcript = await transcriber.transcribe({ mediaFilePath: frVideoPath, language: 'fr', format: 'txt', model: new WhisperBuiltinModel('small') })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'communiquer-lors-dune-classe-transplantee.txt'),
       language: 'fr',

packages/tests/src/transcription/whisper/transcriber/timestamped-transcriber.spec.ts

@@ -4,7 +4,15 @@ import { createLogger } from 'winston'
 import { join } from 'path'
 import { mkdir, rm } from 'node:fs/promises'
 import { buildAbsoluteFixturePath, root } from '@peertube/peertube-node-utils'
-import { OpenaiTranscriber, WhisperTimestampedTranscriber, TranscriptFile, TranscriptFileEvaluator } from '@peertube/peertube-transcription'
+import {
+  OpenaiTranscriber,
+  WhisperTimestampedTranscriber,
+  TranscriptFile,
+  TranscriptFileEvaluator,
+  TranscriptionModel,
+  WhisperTranscribeArgs,
+  WhisperBuiltinModel
+} from '@peertube/peertube-transcription'
 
 config.truncateThreshold = 0
 
@@ -29,15 +37,10 @@ describe('Linto timestamped Whisper transcriber', function () {
   })
 
   it('Should transcribe a media file and produce a transcript file in `vtt` with a ms precision', async function () {
-    const transcript = await transcriber.transcribe(
-      shortVideoPath,
-      { name: 'tiny' },
-      'fr'
-    )
-
+    const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en' })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'video_short.vtt'),
-      language: 'fr',
+      language: 'en',
       format: 'vtt'
     }))).to.be.true
 
@@ -52,7 +55,7 @@ you
   })
 
   it('May produce a transcript file in the `srt` format with a ms precision', async function () {
-    const transcript = await transcriber.transcribe(shortVideoPath, { name: 'tiny' }, 'en', 'srt')
+    const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'srt' })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'video_short.srt'),
       language: 'en',
@@ -69,7 +72,7 @@ you
   })
 
   it('May produce a transcript file in `txt` format', async function () {
-    const transcript = await transcriber.transcribe(shortVideoPath, { name: 'tiny' }, 'en', 'txt')
+    const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'txt' })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'video_short.txt'),
       language: 'en',
@@ -81,12 +84,17 @@ you
   })
 
   it('May transcribe a media file using a local PyTorch model file', async function () {
-    await transcriber.transcribe(frVideoPath, { name: 'myLocalModel', path: buildAbsoluteFixturePath('transcription/models/tiny.pt') }, 'fr')
+    await transcriber.transcribe({ mediaFilePath: frVideoPath, model: TranscriptionModel.fromPath(buildAbsoluteFixturePath('transcription/models/tiny.pt')), language: 'en' })
   })
 
   it('May transcribe a media file in french', async function () {
-    this.timeout(45000)
-    const transcript = await transcriber.transcribe(frVideoPath, { name: 'tiny' }, 'fr', 'txt')
+    this.timeout(2 * 1000 * 60)
+    const transcript = await transcriber.transcribe({
+      mediaFilePath: frVideoPath,
+      language: 'fr',
+      format: 'txt',
+      model: new WhisperBuiltinModel('tiny')
+    })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'communiquer-lors-dune-classe-transplantee.txt'),
       language: 'fr',
@@ -118,13 +126,13 @@ Ensuite, il pourront lire et commenter ce de leur camarade, ou répondre au comm
 
   it('Should produce a text transcript similar to openai-whisper implementation', async function () {
     this.timeout(5 * 1000 * 60)
-    const transcribeArguments: Parameters<typeof transcriber.transcribe> = [
-      frVideoPath,
-      { name: 'tiny' },
-      'fr',
-      'txt'
-    ]
-    const transcript = await transcriber.transcribe(...transcribeArguments)
+    const transcribeArgs: WhisperTranscribeArgs = {
+      mediaFilePath: frVideoPath,
+      model: TranscriptionModel.fromPath(buildAbsoluteFixturePath('transcription/models/tiny.pt')),
+      language: 'fr',
+      format: 'txt'
+    }
+    const transcript = await transcriber.transcribe(transcribeArgs)
 
     const openaiTranscriber = new OpenaiTranscriber(
       {
@@ -137,7 +145,7 @@ Ensuite, il pourront lire et commenter ce de leur camarade, ou répondre au comm
       createLogger(),
       join(transcriptDirectory, 'openai-whisper')
     )
-    const openaiTranscript = await openaiTranscriber.transcribe(...transcribeArguments)
+    const openaiTranscript = await openaiTranscriber.transcribe(transcribeArgs)
 
     const transcriptFileEvaluator = new TranscriptFileEvaluator(openaiTranscript, transcript)
     expect(await transcriptFileEvaluator.wer()).to.be.below(25 / 100)

packages/tests/src/transcription/whisper/transcriber/whisper-ctranslate2.spec.ts

@@ -4,7 +4,13 @@ import { createLogger } from 'winston'
 import { join } from 'path'
 import { mkdir, readFile, rm } from 'node:fs/promises'
 import { buildAbsoluteFixturePath, root } from '@peertube/peertube-node-utils'
-import { Ctranslate2Transcriber, OpenaiTranscriber, TranscriptFile, TranscriptFileEvaluator } from '@peertube/peertube-transcription'
+import {
+  Ctranslate2Transcriber,
+  OpenaiTranscriber,
+  TranscriptFile,
+  TranscriptFileEvaluator,
+  TranscriptionModel, WhisperTranscribeArgs
+} from '@peertube/peertube-transcription'
 
 config.truncateThreshold = 0
 
@@ -29,7 +35,7 @@ describe('Whisper CTranslate2 transcriber', function () {
   })
 
   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(shortVideoPath, { name: 'tiny' })
+    const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en' })
     expect(await transcript.equals(new TranscriptFile({ path: join(transcriptDirectory, 'video_short.vtt') }))).to.be.true
     expect(await readFile(transcript.path, 'utf8')).to.equal(
       `WEBVTT
@@ -42,7 +48,7 @@ You
   })
 
   it('May produce a transcript file in the `srt` format', async function () {
-    const transcript = await transcriber.transcribe(shortVideoPath, { name: 'tiny' }, 'en', 'srt')
+    const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'srt' })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'video_short.srt'),
       format: 'srt'
@@ -58,7 +64,7 @@ You
   })
 
   it('May produce a transcript file in the `txt` format', async function () {
-    const transcript = await transcriber.transcribe(shortVideoPath, { name: 'tiny' }, 'en', 'txt')
+    const transcript = await transcriber.transcribe({ mediaFilePath: shortVideoPath, language: 'en', format: 'txt' })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'video_short.txt'),
       format: 'txt'
@@ -69,12 +75,13 @@ You
   })
 
   it('May transcribe a media file using a local CTranslate2 model', async function () {
-    const transcript = await transcriber.transcribe(
-      shortVideoPath,
-      { name: 'myLocalModel', path: buildAbsoluteFixturePath('transcription/models/faster-whisper-tiny') },
-      'en',
-      'txt'
-    )
+    this.timeout(2 * 1000 * 60)
+    const transcript = await transcriber.transcribe({
+      mediaFilePath: shortVideoPath,
+      model: TranscriptionModel.fromPath(buildAbsoluteFixturePath('transcription/models/faster-whisper-tiny')),
+      language: 'en',
+      format: 'txt'
+    })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'video_short.txt'),
       format: 'txt'
@@ -85,8 +92,8 @@ You
   })
 
   it('May transcribe a media file in french', async function () {
-    this.timeout(45000)
-    const transcript = await transcriber.transcribe(frVideoPath, { name: 'tiny' }, 'fr', 'txt')
+    this.timeout(5 * 1000 * 60)
+    const transcript = await transcriber.transcribe({ mediaFilePath: frVideoPath, language: 'fr', format: 'txt' })
     expect(await transcript.equals(new TranscriptFile({
       path: join(transcriptDirectory, 'communiquer-lors-dune-classe-transplantee.txt'),
       language: 'fr',
@@ -109,13 +116,13 @@ Ensuite, il pourront lire et commenter ce de leur camarade, on répondra au comm
 
   it('Should produce a text transcript similar to openai-whisper implementation', async function () {
     this.timeout(5 * 1000 * 60)
-    const transcribeArguments: Parameters<typeof transcriber.transcribe> = [
-      frVideoPath,
-      { name: 'tiny' },
-      'fr',
-      'txt'
-    ]
-    const transcript = await transcriber.transcribe(...transcribeArguments)
+    const transcribeArgs: WhisperTranscribeArgs = {
+      mediaFilePath: frVideoPath,
+      model: TranscriptionModel.fromPath(buildAbsoluteFixturePath('transcription/models/tiny.pt')),
+      language: 'fr',
+      format: 'txt'
+    }
+    const transcript = await transcriber.transcribe(transcribeArgs)
     const openaiTranscriber = new OpenaiTranscriber(
       {
         name: 'openai-whisper',
@@ -127,7 +134,7 @@ Ensuite, il pourront lire et commenter ce de leur camarade, on répondra au comm
       createLogger(),
       join(transcriptDirectory, 'openai-whisper')
     )
-    const openaiTranscript = await openaiTranscriber.transcribe(...transcribeArguments)
+    const openaiTranscript = await openaiTranscriber.transcribe(transcribeArgs)
 
     const transcriptFileEvaluator = new TranscriptFileEvaluator(openaiTranscript, transcript)
     expect(await transcriptFileEvaluator.wer()).to.be.below(20 / 100)

packages/transcription/README.md

@@ -33,24 +33,26 @@ import { OpenaiTranscriber } from '@peertube/peertube-transcription'
     binary: 'whisper'
   });
 
-  const transcriptFile = await transcriber.transcribe(
-    './myVideo.mp4',
-    { name: 'tiny' },
-    'en', 'txt'
-  );
+  const transcriptFile = await transcriber.transcribe({
+    mediaFilePath: './myVideo.mp4',
+    format: 'txt'
+  });
 
   console.log(transcriptFile.path);
   console.log(await transcriptFile.read());
 })();
 ```
 
 Using a local model file:
+
 ```typescript
-  const transcriptFile = await transcriber.transcribe(
-    './myVideo.mp4',
-    { name: 'my fine tuned large model', path: './models/large.pt' },
-    'en', 'txt'
-  );
+import { WhisperBuiltinModel } from '@peertube/peertube-transcription/dist'
+
+const transcriptFile = await transcriber.transcribe({
+  mediaFilePath: './myVideo.mp4',
+  model: WhisperBuiltinModel.fromPath('./models/large.pt'),
+  format: 'txt'
+});
 ```
 
 You may use the builtin Factory if you're happy with the default configuration:

packages/transcription/src/abstract-transcriber.ts

@@ -9,6 +9,14 @@ import { TranscriptionModel } from './transcription-model.js'
 import { TranscriptionRun } from './transcription-run.js'
 import { TranscriptFile, TranscriptFormat } from './transcript/index.js'
 
+export interface TranscribeArgs {
+  mediaFilePath: string
+  model: TranscriptionModel
+  language?: string
+  format?: TranscriptFormat
+  runId?: SUUID
+}
+
 export abstract class AbstractTranscriber {
   public static DEFAULT_TRANSCRIPT_DIRECTORY = join(root(), 'dist', 'transcripts')
 
@@ -55,11 +63,11 @@ export abstract class AbstractTranscriber {
     return model.format === 'PyTorch'
   }
 
-  abstract transcribe (
-    mediaFilePath: string,
-    model: TranscriptionModel,
-    language: string,
-    format: TranscriptFormat,
-    runId: SUUID
-  ): Promise<TranscriptFile>
+  abstract transcribe ({
+    mediaFilePath,
+    model,
+    language,
+    format = 'vtt',
+    runId = short.generate()
+  }: TranscribeArgs): Promise<TranscriptFile>
 }