@remotion/whisper-web 4.0.364 → 4.0.366

package/src/load-mod/load-mod.ts

@@ -1,19 +0,0 @@
-// ⚠️⚠️⚠️⚠️⚠️!! Intentionally putting this in a subdirectory, so it is 2 directories deep
-// That way it can be imported when the output is dist/esm/index.js
-
-import type {printHandler} from '../print-handler.js';
-
-export const loadMod = async ({
-	handler,
-}: {
-	handler: ReturnType<typeof printHandler>;
-}) => {
-	const createModule = await import('../../main.js').then((mod) => mod.default);
-
-	const Module = await createModule({
-		print: handler,
-		printErr: handler,
-	});
-
-	return Module;
-};

package/src/log.ts

@@ -1,41 +0,0 @@
-/* eslint-disable no-console */
-export const logLevels = ['trace', 'verbose', 'info', 'warn', 'error'] as const;
-
-export type LogLevel = (typeof logLevels)[number];
-
-const getNumberForLogLevel = (level: LogLevel) => {
-	return logLevels.indexOf(level);
-};
-
-export const isEqualOrBelowLogLevel = (
-	currentLevel: LogLevel,
-	level: LogLevel,
-) => {
-	return getNumberForLogLevel(currentLevel) <= getNumberForLogLevel(level);
-};
-
-export const Log = {
-	trace: (logLevel: LogLevel, ...args: Parameters<typeof console.log>) => {
-		if (isEqualOrBelowLogLevel(logLevel, 'trace')) {
-			return console.log(...args);
-		}
-	},
-	verbose: (logLevel: LogLevel, ...args: Parameters<typeof console.log>) => {
-		if (isEqualOrBelowLogLevel(logLevel, 'verbose')) {
-			return console.log(...args);
-		}
-	},
-	info: (logLevel: LogLevel, ...args: Parameters<typeof console.log>) => {
-		if (isEqualOrBelowLogLevel(logLevel, 'info')) {
-			return console.log(...args);
-		}
-	},
-	warn: (logLevel: LogLevel, ...args: Parameters<typeof console.log>) => {
-		if (isEqualOrBelowLogLevel(logLevel, 'warn')) {
-			return console.warn(...args);
-		}
-	},
-	error: (...args: Parameters<typeof console.log>) => {
-		return console.error(...args);
-	},
-};

package/src/mod.ts

@@ -1,13 +0,0 @@
-import type {TranscriptionJson} from './result';
-
-export type ModelState = {
-	transcriptionProgressPlayback: null | ((progress: number) => void);
-	transcriptionChunkPlayback:
-		| null
-		| ((
-				timestampStart: string,
-				timestampEnd: string,
-				textOnly: string,
-		  ) => void);
-	resolver: null | ((transcript: TranscriptionJson) => void);
-};

package/src/print-handler.ts

@@ -1,39 +0,0 @@
-import type {LogLevel} from './log';
-import {Log} from './log';
-import type {TranscriptionJson} from './result';
-
-const RESULT_TOKEN = 'remotion_final:';
-const PROGRESS_TOKEN = 'remotion_progress:';
-const UPDATE_TOKEN = 'remotion_update:';
-const BUSY_TOKEN = 'remotion_busy:';
-
-export const printHandler = ({
-	onProgress,
-	onDone,
-	onBusy,
-	onUpdate,
-	logLevel,
-}: {
-	onProgress: (value: number) => void;
-	onBusy: () => void;
-	onDone: (value: TranscriptionJson) => void;
-	onUpdate: (value: TranscriptionJson) => void;
-	logLevel: LogLevel;
-}) => {
-	return (text: string) => {
-		Log.verbose(logLevel, text);
-
-		if (text.startsWith(PROGRESS_TOKEN)) {
-			const value = parseInt(text.slice(PROGRESS_TOKEN.length), 10);
-			onProgress(value);
-		} else if (text.startsWith(RESULT_TOKEN)) {
-			const json = JSON.parse(text.slice(RESULT_TOKEN.length));
-			onDone(json);
-		} else if (text.startsWith(UPDATE_TOKEN)) {
-			const json = JSON.parse(text.slice(UPDATE_TOKEN.length));
-			onUpdate(json);
-		} else if (text.startsWith(BUSY_TOKEN)) {
-			onBusy();
-		}
-	};
-};

package/src/resample-to-16khz.ts

@@ -1,105 +0,0 @@
-import type {LogLevel} from './log';
-import {Log} from './log';
-
-// this is a requirement of whisper.cpp
-export const EXPECTED_SAMPLE_RATE = 16000;
-
-let context: AudioContext | undefined;
-
-const getAudioContext = () => {
-	if (!context) {
-		context = new AudioContext({
-			sampleRate: EXPECTED_SAMPLE_RATE,
-		});
-	}
-
-	return context;
-};
-
-const audioDecoder = async (
-	audioBuffer: AudioBuffer,
-): Promise<Float32Array> => {
-	const offlineContext = new OfflineAudioContext(
-		audioBuffer.numberOfChannels,
-		audioBuffer.length,
-		audioBuffer.sampleRate,
-	);
-
-	const source = offlineContext.createBufferSource();
-	source.buffer = audioBuffer;
-	source.connect(offlineContext.destination);
-	source.start(0);
-
-	const renderedBuffer = await offlineContext.startRendering();
-	return renderedBuffer.getChannelData(0);
-};
-
-export type ResampleTo16KhzParams = {
-	file: Blob;
-	onProgress?: (p: number) => void;
-	logLevel?: LogLevel;
-};
-
-export const resampleTo16Khz = async ({
-	file,
-	onProgress,
-	logLevel = 'info',
-}: ResampleTo16KhzParams): Promise<Float32Array> => {
-	Log.info(logLevel, `Starting resampling for file, size: ${file.size}`);
-	onProgress?.(0);
-
-	if (typeof window === 'undefined') {
-		Log.error(
-			logLevel,
-			'Window object not found. Resampling can only be done in a browser environment.',
-		);
-		throw new Error(
-			'Window object not found. Resampling requires a browser environment.',
-		);
-	}
-
-	if (!file) {
-		Log.error(logLevel, 'File is empty.');
-		throw new Error('File is empty');
-	}
-
-	const innerContext = getAudioContext();
-	const reader = new FileReader();
-
-	return new Promise<Float32Array>((resolve, reject) => {
-		reader.onprogress = (event) => {
-			if (event.lengthComputable) {
-				const percentage = (event.loaded / event.total) * 0.5; // File reading up to 50%
-				onProgress?.(Math.min(0.5, percentage));
-			}
-		};
-
-		reader.onload = async () => {
-			try {
-				Log.info(logLevel, 'File reading complete. Decoding audio data...');
-				onProgress?.(0.5);
-				const buffer = new Uint8Array(reader.result as ArrayBuffer);
-				const audioBuffer = await innerContext.decodeAudioData(
-					buffer.buffer as ArrayBuffer,
-				);
-				Log.info(logLevel, 'Audio decoding complete. Starting rendering...');
-				onProgress?.(0.75);
-
-				const processedAudio = await audioDecoder(audioBuffer);
-				Log.info(logLevel, 'Audio resampling and processing complete.');
-				onProgress?.(1);
-				resolve(processedAudio);
-			} catch (error) {
-				Log.error(logLevel, 'Error during audio processing:', error);
-				reject(error);
-			}
-		};
-
-		reader.onerror = () => {
-			Log.error(logLevel, 'File reading failed.');
-			reject(new Error('File reading failed'));
-		};
-
-		reader.readAsArrayBuffer(file);
-	});
-};

package/src/result.ts

@@ -1,59 +0,0 @@
-type Result = {
-	language: string;
-};
-
-type Model = {
-	type: string;
-	multilingual: boolean;
-	vocab: number;
-	audio: {
-		ctx: number;
-		state: number;
-		head: number;
-		layer: number;
-	};
-	text: {
-		ctx: number;
-		state: number;
-		head: number;
-		layer: number;
-	};
-	mels: number;
-	ftype: number;
-};
-
-type Timestamps = {
-	from: string;
-	to: string;
-};
-
-type Offsets = {
-	from: number;
-	to: number;
-};
-
-type TranscriptionItem = {
-	timestamps: Timestamps;
-	offsets: Offsets;
-	text: string;
-};
-
-export type WordLevelToken = {
-	t_dtw: number;
-	text: string;
-	timestamps: Timestamps;
-	offsets: Offsets;
-	id: number;
-	p: number;
-};
-
-export type TranscriptionItemWithTimestamp = TranscriptionItem & {
-	tokens: WordLevelToken[];
-};
-
-export type TranscriptionJson = {
-	systeminfo: string;
-	model: Model;
-	result: Result;
-	transcription: TranscriptionItemWithTimestamp[];
-};

package/src/simulate-progress.ts

@@ -1,74 +0,0 @@
-import {
-	getActualTranscriptionSpeedInMilliseconds,
-	NEW_PROGRESS_EVENT_EVERY_N_SECONDS,
-	storeActualTranscriptionSpeed,
-} from './transcription-speed';
-
-export const simulateProgress = ({
-	audioDurationInSeconds,
-	onProgress,
-}: {
-	audioDurationInSeconds: number;
-	onProgress: (num: number) => void;
-}) => {
-	let progress = 0;
-
-	const progressSteps =
-		audioDurationInSeconds / NEW_PROGRESS_EVENT_EVERY_N_SECONDS;
-
-	let progressStepsReceived = 0;
-	let timer: NodeJS.Timer | null = null;
-	let lastTimerStart: number | null = null;
-
-	const start = () => {
-		const speed = getActualTranscriptionSpeedInMilliseconds();
-
-		let iterations = 0;
-		lastTimerStart = Date.now();
-
-		timer = setInterval(() => {
-			progress += 1 / NEW_PROGRESS_EVENT_EVERY_N_SECONDS / (progressSteps + 1);
-			progress = Math.min(progress, 0.99);
-			onProgress(progress);
-			iterations += 1;
-			if (iterations > NEW_PROGRESS_EVENT_EVERY_N_SECONDS - 1 && timer) {
-				clearInterval(timer);
-				timer = null;
-			}
-		}, speed / NEW_PROGRESS_EVENT_EVERY_N_SECONDS);
-	};
-
-	return {
-		start,
-		progressStepReceived: () => {
-			progressStepsReceived += 1;
-			progress = progressStepsReceived / progressSteps;
-			if (timer) {
-				clearInterval(timer);
-				timer = null;
-			}
-
-			if (lastTimerStart) {
-				const timeToProcessChunk = Date.now() - (lastTimerStart ?? Date.now());
-				storeActualTranscriptionSpeed(timeToProcessChunk);
-			}
-
-			start();
-		},
-		onDone: () => {
-			if (timer) {
-				clearInterval(timer);
-				timer = null;
-			}
-
-			progress = 1;
-			onProgress(1);
-		},
-		abort: () => {
-			if (timer) {
-				clearInterval(timer);
-				timer = null;
-			}
-		},
-	};
-};

package/src/to-captions.ts

@@ -1,37 +0,0 @@
-import type {Caption} from '@remotion/captions';
-import type {TranscriptionItemWithTimestamp, TranscriptionJson} from './result';
-
-type ToCaptionsInput = {
-	whisperWebOutput: TranscriptionJson | TranscriptionItemWithTimestamp[];
-};
-
-type ToCaptionsOutput = {
-	captions: Caption[];
-};
-
-export const toCaptions = (input: ToCaptionsInput): ToCaptionsOutput => {
-	const transcription =
-		'transcription' in input.whisperWebOutput
-			? input.whisperWebOutput.transcription
-			: input.whisperWebOutput;
-
-	const captions: Caption[] = [];
-
-	for (const item of transcription) {
-		if (item.text === '') {
-			continue;
-		}
-
-		for (const token of item.tokens) {
-			captions.push({
-				text: captions.length === 0 ? token.text.trimStart() : token.text,
-				startMs: token.offsets.from,
-				endMs: token.offsets.to,
-				timestampMs: token.t_dtw === -1 ? null : token.t_dtw * 10,
-				confidence: token.p,
-			});
-		}
-	}
-
-	return {captions};
-};

package/src/transcribe.ts

@@ -1,173 +0,0 @@
-/* eslint-disable new-cap */
-import type {MainModule} from '../main';
-import type {WhisperWebLanguage, WhisperWebModel} from './constants';
-import {getObject} from './db/get-object-from-db';
-import {getModelUrl} from './get-model-url';
-import {loadMod} from './load-mod/load-mod';
-import type {LogLevel} from './log';
-import {Log} from './log';
-import {printHandler} from './print-handler';
-import {EXPECTED_SAMPLE_RATE} from './resample-to-16khz';
-import type {TranscriptionItemWithTimestamp, TranscriptionJson} from './result';
-import {simulateProgress} from './simulate-progress';
-const MAX_THREADS_ALLOWED = 16;
-const DEFAULT_THREADS = 4;
-
-declare global {
-	interface Window {
-		remotion_wasm_moduleOverrides?: Record<string, (...args: any[]) => void>;
-	}
-}
-
-interface WithResolvers<T> {
-	promise: Promise<T>;
-	resolve: (value: T | PromiseLike<T>) => void;
-	// eslint-disable-next-line @typescript-eslint/no-explicit-any
-	reject: (reason?: any) => void;
-}
-
-const withResolvers = function <T>() {
-	let resolve: WithResolvers<T>['resolve'];
-	let reject: WithResolvers<T>['reject'];
-	const promise = new Promise<T>((res, rej) => {
-		resolve = res;
-		reject = rej;
-	});
-	return {promise, resolve: resolve!, reject: reject!};
-};
-
-export type TranscribeParams = {
-	channelWaveform: Float32Array;
-	model: WhisperWebModel;
-	language?: WhisperWebLanguage;
-	onProgress?: (p: number) => void;
-	onTranscriptionChunk?: (
-		transcription: TranscriptionItemWithTimestamp[],
-	) => void;
-	threads?: number;
-	logLevel?: LogLevel;
-};
-
-const storeFS = (mod: MainModule, fname: string, buf: any) => {
-	try {
-		mod.FS_unlink(fname);
-	} catch {
-		// ignore
-	}
-
-	mod.FS_createDataFile('/', fname, buf, true, true, undefined);
-};
-
-export const transcribe = async ({
-	channelWaveform,
-	model,
-	language = 'auto',
-	onProgress,
-	threads,
-	onTranscriptionChunk,
-	logLevel = 'info',
-}: TranscribeParams): Promise<TranscriptionJson> => {
-	if (!channelWaveform || channelWaveform.length === 0) {
-		Log.error(logLevel, 'No audio data provided or audio data is empty.');
-		throw new Error('No audio data provided or audio data is empty.');
-	}
-
-	Log.info(
-		logLevel,
-		`Starting transcription with model: ${model}, language: ${language}, threads: ${
-			threads ?? DEFAULT_THREADS
-		}`,
-	);
-
-	if ((threads ?? DEFAULT_THREADS) > MAX_THREADS_ALLOWED) {
-		Log.warn(
-			logLevel,
-			`Thread limit exceeded: Used ${
-				threads ?? DEFAULT_THREADS
-			}, max ${MAX_THREADS_ALLOWED} allowed.`,
-		);
-		return Promise.reject(
-			new Error(`Thread limit exceeded: max ${MAX_THREADS_ALLOWED} allowed.`),
-		);
-	}
-
-	const audioDurationInSeconds = channelWaveform.length / EXPECTED_SAMPLE_RATE;
-
-	const {
-		abort: abortProgress,
-		onDone: onProgressDone,
-		progressStepReceived,
-		start: startProgress,
-	} = simulateProgress({
-		audioDurationInSeconds,
-		onProgress: (p) => {
-			onProgress?.(p);
-		},
-	});
-
-	const {
-		promise,
-		resolve: _resolve,
-		reject: _reject,
-	} = withResolvers<TranscriptionJson>();
-
-	const resolve = (value: TranscriptionJson) => {
-		_resolve(value);
-		abortProgress();
-		Log.info(logLevel, 'Transcription completed successfully.');
-	};
-
-	const reject = (reason: Error) => {
-		_reject(reason);
-		abortProgress();
-		Log.error('Transcription failed:', reason);
-	};
-
-	const handler = printHandler({
-		logLevel,
-		onProgress: (p: number) => {
-			if (p === 0) {
-				startProgress();
-			} else if (p === 100) {
-				onProgressDone();
-			} else {
-				progressStepReceived();
-			}
-		},
-		onDone: resolve,
-		onBusy: () => {
-			reject(new Error('Another transcription is already in progress'));
-		},
-		onUpdate: (json: TranscriptionJson) => {
-			onTranscriptionChunk?.(json.transcription);
-		},
-	});
-
-	const Mod = await loadMod({handler});
-
-	const url = getModelUrl(model);
-	const result = await getObject({key: url});
-	if (!result) {
-		throw new Error(
-			`Model ${model} is not loaded. Call downloadWhisperModel() first.`,
-		);
-	}
-
-	Log.info(logLevel, `Model ${model} loaded successfully.`);
-
-	const fileName = `${model}.bin`;
-
-	storeFS(Mod, fileName, result);
-
-	Log.info(logLevel, 'Starting main transcription process...');
-	Mod.full_default(
-		fileName,
-		channelWaveform,
-		model,
-		language,
-		threads ?? DEFAULT_THREADS,
-		false,
-	);
-
-	return promise;
-};

package/src/transcription-speed.ts

@@ -1,21 +0,0 @@
-export const storeActualTranscriptionSpeed = (speed: number) => {
-	window.localStorage.setItem(
-		'remotion-whisper-web-transcription-speed',
-		speed.toString(),
-	);
-};
-
-// conservative estimate that 30 seconds of audio takes 30 second to process
-const DEFAULT_ASSUMED_SPEED = 1;
-export const NEW_PROGRESS_EVENT_EVERY_N_SECONDS = 30;
-
-export const getActualTranscriptionSpeedInMilliseconds = () => {
-	const speed = window.localStorage.getItem(
-		'remotion-whisper-web-transcription-speed',
-	);
-	if (!speed) {
-		return DEFAULT_ASSUMED_SPEED * NEW_PROGRESS_EVENT_EVERY_N_SECONDS * 1000;
-	}
-
-	return parseFloat(speed);
-};

package/tsconfig.json

@@ -1,10 +0,0 @@
-{
-	"extends": "../tsconfig.settings.json",
-	"compilerOptions": {
-		"rootDir": "src",
-		"outDir": "dist"
-	},
-	"include": ["src", "main.d.ts"],
-	"exclude": ["src/index.module.ts"],
-	"references": [{"path": "../captions"}]
-}