Demos
Speech Demo (Android)
Echo pipeline demo — transcribe speech and synthesize it back. Real-time VAD waveform visualization, chat bubble UI with STT/TTS latency display.
Android 8+Parakeet TDTKokoroSilero VADDeepFilterNet3
Linux CLI Demo
Command-line speech pipeline for embedded Linux. ALSA mic input or stdin PCM pipe.
Linux ARM64Linux x86_64ONNX Runtime
Android Demo
Download the pre-built APK (20 MB) or build from source:
app-release.apk (Signed APK, 20 MB)- Models auto-download from HuggingFace (~1.2 GB on first launch)
- Parakeet TDT v3 STT (114 languages), Kokoro 82M TTS (7 languages)
- Silero VAD, DeepFilterNet3 noise cancellation, barge-in support
Linux CLI Demo
Build and run:
cd linux && cmake -B build && cmake --build build
./build/speech_demo --model-dir /path/to/models