wzhd / vosk-rs

Cloud-free speech recognition. See https://fars.ee/F9-b.mp4

Vosk

Vosk is a speech recognition toolkit with state-of-the-art accuracy and low latency. It runs locally and recognises each word as it is uttered without waiting for the end of a sentence.

Usage

Get (or train) a model. There are trained models for 16 languages ready for downloading.

Load the model from disk.

let model = Model::new("path_to_model_directory").unwrap();

Create a recognizer object, set the sample rate of audio data.

let mut recognizer = Recognizer::new(&model, 16000.0);

Feed 16-bit audio data as an i16 slice.

recognizer.accept_waveform(input);

Use recognizer.partial_result() to get the words being uttered. Use recognizer.result() to get completed sentences and additional information such as timing and confidence scores.

Dependency

Build vosk-sys.

vosk-rs's Issues

Using with Cargo

Hello! Is this crate going to be uploaded to crates.io anytime soon? I'd like to use it and it's handier to do it that way.

License

Hi, would you be interested in adding a license to this project as well as vosk-sys?