@environment-safe/speech v0.0.6

@environment-safe/speech

Text to Speech and speech recognition in node and the browser. Supports native STT/TTS where possible, falls back to local support or native browser support where appropriate.

Usage

In node this looks like:

import { Speech } from '@environment-safe/speech';
(async ()=>{
    await Speech.ready;
    const voice = Speech.voices[0].name;
    await Speech.speak('I have something to say', { voice });
})();
// node.js STT support is still TBD

In the browser, this must occur on an execution context initiated by a user action, like:

import { Speech } from '@environment-safe/speech';
const el = document.getElementById('element-id');
el.addEventListener('click', async ()=>{
    const spoken = await Speech.hear();
    //do something
});
//OR
el.addEventListener('click', async ()=>{
    const voice = Speech.voices[0].name;
    await Speech.speak('I have something to say', { voice });
    //do something
});

=== Engine Type Support

Supported STT engines are: browser, sherpa-onnx and whisper. Supported TTS engines are: browser, mac-os-x, windows, node, and sherpa-onnx. Autodetection prefers native, local solutions.

Outside the browser getting command line access to native STT is difficult, so the STT options in node are currently bulky and limited. This should change over time.

Testing

Run the es module tests to test the root modules

npm run import-test

Because we don't currently support wantsInput on the Speech API you have to run the demo to test in the browser. To do that:

npm run local-server

Development

All work is done in the .mjs files and will be transpiled on commit to commonjs and tested.

If the above tests pass, then attempt a commit which will generate .d.ts files alongside the src files and commonjs classes in dist