Crate wavekat_asr 

wavekat-asr

Streaming ASR trait surface, intended to wrap one or more speech-to-text backends behind a common Rust API. Modeled on the same pattern as wavekat-vad and wavekat-turn.

Status

This crate is pre-1.0. The trait surface may iterate as more backends land. Pin to an exact patch version.

The bundled backend is backends::sherpa_onnx (behind the sherpa-onnx Cargo feature): a local streaming Zipformer that auto-downloads its model from HuggingFace on first use.

Re-exports

pub use error::AsrError;

Modules

backends

Backend implementations of crate::StreamingAsr.

error

Structs

AudioFrame

A frame of audio samples with associated sample rate.

Enums

Channel

Which side of a two-channel call the audio (or transcript) belongs to.

TranscriptEvent

One transcript event emitted by a StreamingAsr backend.

Traits

StreamingAsr

A streaming ASR session.