Akiki

Run streaming voice activity detection locally with a small bundled Akiki VAD model. No model download, network connection or PyTorch is needed at runtime.

pip install akiki
akiki --version
akiki recording.wav

Audio must be mono at 16,000 Hz. The CLI prints one JSON object per complete 16 ms frame: its start, end, speech probability and threshold decision.

import soundfile as sf
from akiki import VAD

audio, sample_rate = sf.read("recording.wav", dtype="float32")
vad = VAD()
for frame in vad.push(audio, sample_rate=sample_rate):
    print(frame)
vad.reset()  # Start a new recording.

push() accepts arbitrary buffer sizes and retains incomplete frames between calls. A trailing incomplete frame is left pending; it is not padded or scored. Use VAD(threshold=0.6) or akiki recording.wav --threshold 0.6 to change the threshold. Probabilities are always returned.

Version 0.0.1 bundles the experimental candidate-1, seed 1729 model, with 210,561 parameters and a 16 ms output hop. It is an initial functional release; future versions will supply the production Akiki VAD models. Training provenance and checksums are included in akiki/models/MODEL_CARD.md and the model metadata.

Code and bundled Akiki weights are provided under Apache-2.0. Source audio and the independent teacher retain their original licences; neither is distributed in this package.

To publish this release from the project directory:

uv build
uv publish dist/akiki-0.0.1*

Authenticate to PyPI locally when prompted. The project contains no credentials.