Flutter TensorFlow Speech

pub package

The TensorFlow audio recognition tutorial, for use in flutter, with an API that you'll love to use!

var speech = TfSpeech();
await for (var result in speech.stream) {

Example app video

Source code.

How does this work?

We use Android's AudioRecord API to record audio in the smallest possible chunks.

These chunks are loaded into a ring buffer.

The ring buffer is periodically fed into the TensorFlow model for inference.

The raw output from the model is passed straight to dart, which allows for a great degree of control from dart code.