MLX-Audio
An audio-processing library built on Apple's MLX framework for fast text-to-speech, speech-to-text, speech-to-speech and music generation on Apple Silicon.
MLX-Audio is an audio-processing library built on Apple's MLX framework, providing text-to-speech, speech-to-text, speech-to-speech and music generation with inference optimized for Apple Silicon's M-series chips.
It supports multiple model architectures across these tasks, multilingual voices, voice customization and cloning, and adjustable speech speed, alongside an interactive web interface with 3D audio visualization and an OpenAI-compatible REST API. Model weights can be quantized to 3, 4, 6 or 8 bits for smaller and faster inference, and a Swift package integrates it into iOS and macOS apps.
It installs with pip or uv, and can be used from the command line or from a Python API that loads a model and generates audio from text.