Soniqo

Open-source speech AI that runs on-device.

Soniqo builds local speech models and runtimes for Apple Silicon, iOS, Android, Linux, and Windows: ASR, TTS and voice cloning, VAD, diarization, speech enhancement, and voice-agent pipelines.

Website & docs · GitHub · All models

Model collections

• ONNX — cross-platform and Android model bundles
• LiteRT — efficient on-device models for Android, Linux, and Windows
• MLX Speech Models — optimized for Apple Silicon
• CoreML Speech Models — optimized for Apple Neural Engine on iOS and macOS

Repositories

speech-swift · speech-core · speech-android · speech-studio · runner