MLX-Audio
Speech synthesis, transcription and voice conversion on Apple Silicon
What it is
A Python library and command-line toolkit that runs text-to-speech, speech-to-text, speech-to-speech and music generation models locally on Apple Silicon using the MLX framework. It supports many open models such as Kokoro, Qwen3-TTS, Dia and Chatterbox, with voice cloning, speed control, streaming output and quantization down to a few bits. Developers building voice features for Mac or iOS apps can also use its Swift package, browser interface, or OpenAI-compatible REST API server.