Changelog

Changelog

[3.0.0] - 2025-05-16

Breaking

• removed workerPath & locateFile because there is no separate worker file anymore
  • this is only a breaking changed if you've used these options to locate the separate worker file

Changed

• updated whisper.cpp to v1.7.5
• updated dev dependencies
• disable console.log output in vitest (npm run test:unit)

Fixed

• build with latest emscripten version to fix Vite is unable to parse the worker options; issue #14

[2.0.6] - 2025-02-18

Add

• add params and explanation for preRecordMs, maxRecordMs, minSilenceMs in stream example

Changed

• updated whisper.cpp to v1.7.4
• add integration guide links and updated install instructions in readme
• changed StreamTranscriber.transcribe default params for better results

[2.0.5] - 2024-12-12

Changed

• FileTranscriber.transcribe() now only runs if wasm module and model file are loaded successfully

Fixed

• prevent multiple calls to createModule if model file was not found - fixes possible memory issue #10

[2.0.4] - 2024-11-27

Add

• command to check generated types in packages npm run check-types

Changed

• StreamTranscriber.getAudioWorkletPath() second parameter is now URL|string instead of import.meta

Fixed

• Typescript compile error #8

[2.0.3] - 2024-11-19

Add

• script to update the version string (aka. cache busting) in the example files

Changed

• explicit set wasm build environment to web,webview,webworker, this removes unused code for node.js (which is not supported by transcribe.js)

Fixed

• module import with bundler (eg. SvelteKit); issue #6

[2.0.2] - 2024-11-15

Changed

• wasm memory
  • reduced initial memory allocation to 512MB
  • enabled allow memory growth
• updated whisper.cpp to v1.7.1
• bump @transcribe/shout peer dependency to v1.0.3
• updated dev dependencies

Fixed

• fix Uncaught RangeError: ... could not allocate memory on init() on machines with less memory available

[2.0.1] - 2024-05-31

Changed

• added index.d.ts to @transcribe/shout

[2.0.0] - 2024-05-31

Breaking

Removed the direct dependency of @transcribe/shout in Transcriber classes and pass createModule() as a constructor parameter instead.

Why? The wasm build can't be bundled so a manual import is easier than dealing with bundler exclude configs and remapping imports.

You need to change the following in your code:

+ import createModule from "/your/project/shout.wasm.js";
import { FileTranscriber } from "@transcribe/transcriber";

// create new instance
const transcriber = new FileTranscriber({
+ createModule, // create module function from emscripten build
  model: "/your/project/ggml-tiny-q5_1.bin", // path to ggml model file
  workerPath: "/your/project", // directory of shout.wasm.worker.mjs copied before
});

Changed

• updated whisper.cpp to v1.6.2

Fixed

• align demo styles with docs

[1.0.0] - 2024-05-15

Breaking

• added token_timestamps param to Module.transcribe(...), throws parameter count error if omited
• renamed type TranscripeResult => TranscribeResult

Added

• Transcriber.isReady property
• callback setter in FileTranscriber & StreamTranscriber
  • FileTranscriber.onComplete
  • FileTranscriber.onCanceled
  • FileTranscriber.onProgress
  • FileTranscriber.onSegment
  • StreamTranscriber.onSegement
  • StreamTranscriber.onStreamStatus
• FileTranscriber.destroy() & StreamTranscriber.destroy()
• token_timestamps option to FileTranscriber.transcribe('myfile.mp', {..., token_timestamps: false})

Changed

• type TranscripeResult => TranscribeResult

Fixed

• JSDoc props from private to protected in base class

[0.1.3]

• initial release