@remotion/whisper-webgpu 4.0.522 → 4.0.523

package/dist/index.d.ts

@@ -1,15 +1,17 @@
-export { canUseWhisperWebGpu, WhisperWebGpuUnsupportedReason, } from './can-use-whisper-webgpu';
+export { WhisperWebGpuUnsupportedReason, canUseWhisperWebGpu, } from './can-use-whisper-webgpu';
 export type { CanUseWhisperWebGpuResult } from './can-use-whisper-webgpu';
 export { clearStaleModels } from './clear-stale-models';
-export { disposeWhisperModel, loadWhisperModel } from './load-whisper-model';
-export type { DisposeWhisperModelOptions, LoadWhisperModelOptions, LoadWhisperModelResult, OnWhisperWebGpuModelLoadProgress, WhisperWebGpuModelLoadProgress, } from './load-whisper-model';
 export { isWhisperModelCached } from './is-whisper-model-cached';
 export type { IsWhisperModelCachedOptions } from './is-whisper-model-cached';
-export { getAvailableModels, WHISPER_WEBGPU_MODELS } from './models';
+export { disposeWhisperModel, loadWhisperModel } from './load-whisper-model';
+export type { DisposeWhisperModelOptions, LoadWhisperModelOptions, LoadWhisperModelResult, OnWhisperWebGpuModelLoadProgress, WhisperWebGpuModelLoadProgress, } from './load-whisper-model';
+export { WHISPER_WEBGPU_MODELS, getAvailableModels } from './models';
 export type { WhisperWebGpuModel, WhisperWebGpuModelInfo } from './models';
-export { resampleTo16Khz, WHISPER_WEBGPU_SAMPLE_RATE } from './resample-to-16khz';
+export { removeWhisperModel } from './remove-whisper-model';
+export type { RemoveWhisperModelOptions } from './remove-whisper-model';
+export { WHISPER_WEBGPU_SAMPLE_RATE, resampleTo16Khz } from './resample-to-16khz';
 export type { ResampleTo16KhzOptions } from './resample-to-16khz';
 export { toCaptions } from './to-captions';
 export type { ToCaptionsOptions, ToCaptionsResult } from './to-captions';
 export { transcribe } from './transcribe';
-export type { TranscribeOptions, WhisperWebGpuTranscription, WhisperWebGpuWord, } from './transcribe';
+export type { TranscribeOptions, WhisperWebGpuTask, WhisperWebGpuTranscription, WhisperWebGpuWord, } from './transcribe';

package/dist/models.d.ts

@@ -5,11 +5,16 @@ export type WhisperWebGpuModelInfo = {
     modelId: string;
     parameters: number;
     multilingual: boolean;
+    supportsTranslation: boolean;
     webGpuDownloadSize: number;
 };
 export declare const getAvailableModels: () => WhisperWebGpuModelInfo[];
 export declare const getModelInfo: (model: "base" | "base.en" | "large-v3-turbo" | "medium" | "medium.en" | "small" | "small.en" | "tiny" | "tiny.en") => WhisperWebGpuModelInfo;
-export declare const WHISPER_WEBGPU_DTYPE: {
+export declare const getHostedModelId: (model: "base" | "base.en" | "large-v3-turbo" | "medium" | "medium.en" | "small" | "small.en" | "tiny" | "tiny.en") => string;
+export declare const getWhisperWebGpuDtype: (model: "base" | "base.en" | "large-v3-turbo" | "medium" | "medium.en" | "small" | "small.en" | "tiny" | "tiny.en") => {
     readonly encoder_model: "fp32";
     readonly decoder_model_merged: "q4";
+} | {
+    readonly encoder_model: "fp16";
+    readonly decoder_model_merged: "q4";
 };

package/dist/private-transformers.d.ts

@@ -0,0 +1 @@
+export * from '@huggingface/transformers';

package/dist/remove-whisper-model.d.ts

@@ -0,0 +1,5 @@
+import { type WhisperWebGpuModel } from './models';
+export type RemoveWhisperModelOptions = {
+    model: WhisperWebGpuModel;
+};
+export declare const removeWhisperModel: ({ model, }: RemoveWhisperModelOptions) => Promise<void>;

package/dist/transcribe.d.ts

@@ -10,12 +10,20 @@ export type WhisperWebGpuTranscription = {
     words: WhisperWebGpuWord[];
     model: WhisperWebGpuModel;
 };
+export type WhisperWebGpuTask = 'transcribe' | 'translate';
 export type TranscribeOptions = {
     channelWaveform: Float32Array;
     model: WhisperWebGpuModel;
     language?: string;
+    task?: WhisperWebGpuTask;
     chunkLengthInSeconds?: number;
     strideLengthInSeconds?: number;
+    forceFullSequences?: boolean;
+    doSample?: boolean;
+    temperature?: number;
+    topK?: number;
+    repetitionPenalty?: number;
+    noRepeatNgramSize?: number;
     onModelLoadProgress?: OnWhisperWebGpuModelLoadProgress;
 };
-export declare const transcribe: ({ channelWaveform, model, language, chunkLengthInSeconds, strideLengthInSeconds, onModelLoadProgress, }: TranscribeOptions) => Promise<WhisperWebGpuTranscription>;
+export declare const transcribe: ({ channelWaveform, model, language, task, chunkLengthInSeconds, strideLengthInSeconds, forceFullSequences, doSample, temperature, topK, repetitionPenalty, noRepeatNgramSize, onModelLoadProgress, }: TranscribeOptions) => Promise<WhisperWebGpuTranscription>;

package/dist/with-remotion-model-host.d.ts

@@ -0,0 +1,8 @@
+import type { env, ModelRegistry, pipeline } from '@huggingface/transformers';
+type TransformersModule = {
+    readonly env: typeof env;
+    readonly ModelRegistry: typeof ModelRegistry;
+    readonly pipeline: typeof pipeline;
+};
+export declare const withRemotionModelHost: <ReturnValue>(operation: (transformers: TransformersModule) => Promise<ReturnValue>) => Promise<ReturnValue>;
+export {};

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@remotion/whisper-webgpu",
-	"version": "4.0.522",
+	"version": "4.0.523",
 	"description": "Transcribe audio in the browser using Whisper and WebGPU",
 	"main": "dist/index.js",
 	"types": "dist/index.d.ts",
@@ -24,8 +24,10 @@
 		"url": "https://github.com/remotion-dev/remotion/issues"
 	},
 	"dependencies": {
-		"@huggingface/transformers": "4.2.0",
-		"@remotion/captions": "4.0.522"
+		"@remotion/captions": "4.0.523"
+	},
+	"peerDependencies": {
+		"@huggingface/transformers": "4.2.0"
 	},
 	"exports": {
 		".": {
@@ -36,7 +38,8 @@
 		"./package.json": "./package.json"
 	},
 	"devDependencies": {
-		"@remotion/eslint-config-internal": "4.0.522",
+		"@huggingface/transformers": "4.2.0",
+		"@remotion/eslint-config-internal": "4.0.523",
 		"@typescript/native-preview": "7.0.0-dev.20260217.1",
 		"eslint": "9.19.0"
 	},