A research repo for the WaveKat project — interactive tools and Jupyter notebooks for working with audio models (VAD, turn detection, voice datasets, and more).

wavekat-lab/
├── tools/
│   ├── audio-lab/     Real-time VAD + Turn Detection comparison app (Rust + React)
│   └── cv-explorer/   Mozilla Common Voice dataset browser (Cloudflare Workers + React)
├── notebooks/         Jupyter notebooks (training, validation, dataset splits)
└── docs/              Plans and design docs

Each tool is self-contained — its own Makefile, lockfiles, and build setup live inside its folder.

Web app for testing and comparing WaveKat library backends side by side in real time. Live mic capture, WAV upload, multi-config fan-out, VAD-gated pipeline mode, waveform + spectrogram + probability timelines.

Backends: webrtc-vad, silero-vad, ten-vad, firered-vad, pipecat smart-turn. Details →

Web app for browsing and playing audio clips from the Mozilla Common Voice dataset. Filter by locale, split, demographics, and search sentences — with waveform playback powered by WaveSurfer.js. Built on Cloudflare Workers + D1 + R2. Details →

Live: https://commonvoice-explorer.wavekat.com/

notebooks/ is the home for Jupyter notebooks covering training, validation, and dataset-splitting workflows. Python env is managed by uv.

make setup-notebooks   # one-time: uv sync the notebook env
make lab               # start Jupyter Lab on notebooks/

• Per-tool Makefiles — tools/<name>/Makefile owns dev/build/CI for that tool. Run cd tools/<name> && make help to see what's there.
• Root Makefile — repo-wide only: setup, lab, ci, and per-tool CI delegators.
• No shared Cargo workspace at root — each Rust tool keeps its own Cargo.toml / Cargo.lock / target/ inside its folder.

Video	Description
	Exploring Mozilla Common Voice with Common Voice Explorer Introducing Common Voice Explorer — browse and listen to 1.8M+ real voice clips from the Mozilla Common Voice dataset.
	Testing Pipecat Smart Turn with WaveKat Lab Visual test of Pipecat Smart Turn v3 — live recording and VAD-gated pipeline mode simulating production workflows.
	Adding FireRedVAD as the 4th backend Benchmarking Xiaohongshu's FireRedVAD against Silero, TEN VAD, and WebRTC across accuracy and latency.
	VAD Lab: Real-time multi-backend comparison Live demo of VAD Lab comparing WebRTC, Silero, and TEN VAD side by side with real-time waveform visualization.

Licensed under Apache 2.0.

Copyright 2026 WaveKat.