https://github.com/rxtoolkit/stt-deepspeech

👂 RxJS operator for speech-to-text using DeepSpeech
https://github.com/rxtoolkit/stt-deepspeech

fp functional-programming observables package reactive-programming rxjs speech-to-text stt transcription

Last synced: 5 days ago
JSON representation

👂 RxJS operator for speech-to-text using DeepSpeech

• Host: GitHub
• URL: https://github.com/rxtoolkit/stt-deepspeech
• Owner: rxtoolkit
• License: mit
• Created: 2021-02-09T19:33:25.000Z (almost 4 years ago)
• Default Branch: master
• Last Pushed: 2024-02-16T18:39:26.000Z (11 months ago)
• Last Synced: 2024-11-09T17:36:21.542Z (2 months ago)
• Topics: fp, functional-programming, observables, package, reactive-programming, rxjs, speech-to-text, stt, transcription
• Language: JavaScript
• Homepage:
• Size: 3.45 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 10
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.md

Awesome Lists containing this project

README

        
# @rxtk/toDeepSpeech

> 👂 An RxJS operator for real-time speech-to-text (STT/S2T) streaming using the opensource DeepSpeech library.

```bash

npm i @rxtk/stt-deepspeech

```

```bash

yarn add @rxtk/stt-deepspeech

```

⚠️ To run the DeepSpeech pipeline, you must [download the corresponding DeepSpeech model](https://github.com/mozilla/DeepSpeech/releases), unzip it and pass the model directory to the `toDeepSpeech` operator like this: `toDeepSpeech({modelDir: 'path/to/deepseech-models-0.7.0'})`.

> ⚠️ node.js only. This has not been tested on Browsers but it might be possible to make it work.  If you get it working, please make a PR!

## API

### `toDeepSpeech`

Stream audio speech data to DeepSpeech and get transcripts back:

```js

import {map} from 'rxjs/operators';

import {toDeepSpeech} from '@rxtk/stt-deepspeech';

// The pipeline takes a stream of audio chunks encoded as LINEAR16 (PCM encoded as 16-bit integers) (Buffer, String, Blob or Typed Array)

const buffer$ = pcmChunkEncodedAs16BitIntegers$.pipe(

  map(chunk => Buffer.from(chunk, 'base64')),

  toDeepSpeech({modelDir: '/path/to/deepspeech-models-0.7.0'})

);

buffer$.subscribe(console.log); // log transcript output

```

> ⚠️ Pay attention to the endcoding of the audio data.  The operator only accepts PCM data encoded as 16-bit integers. For example, LINEAR16 encoding usually works.

## Guides

- [Introduction to audio data](https://developer.mozilla.org/en-US/docs/Web/Media/Formats/Audio_concepts)