citrinet-rs 0.1.0

Minimal Citrinet speech-to-text inference using ONNX Runtime and kaldi-native-fbank in Rust.
Documentation

citrinet-rs

Minimal Citrinet inference example in Rust using ONNX Runtime (ort) and kaldi-native-fbank for feature extraction.

Run

cargo run --release -- \
  --model model.onnx \
  --tokens tokens.txt \
  --audio test_wavs/0.wav

Inputs are expected to be mono 16 kHz WAV audio and a token file formatted as symbol index per line (batch size 1). Defaults match the sample paths above.