Speech-To-Text- / Audio-Module-Demo

[mrcwbr]

Area Select

Website/Documentation

Description

Hey I'm trying to use PyTorch live for a Speech-To-Text feature. It would be great if there were a tutorial for this use case.

I read the documentation for the audio module. but I thing the documentation isn't up to date. the module provides only a record function in version "react-native-pytorch-core": "0.1.3", not start and stop-recording as described in the article.

I tried a small playground but the following code leads to an error:

const audio = await AudioUtil.record(2)
console.log('audio', audio)
const {result} = await MobileModel.execute(model_wav2vec, { audio })
console.log('result', result)

there is also a great tutorial for android but none for PyTorch live (react native)

[jeffxtang]

Thanks for your kind note on the Android demo tutorial - glad you found it helpful :)

After you run:

cd live/react-native-pytorch-core/example
npx torchlive-cli run-android

Do you not see the Wav2Vec2 example?

You can also launch this from iOS:
···
cd live/react-native-pytorch-core/example/ios
pod install
open PyTorchCoreExample.xcworkspace
···

The code that handles recording is here.