Kokoro TTS
Kokoro-FastAPI is a Docker-based FastAPI wrapper, specially designed for the Kokoro-82M text-to-speech (TTS) model. It supports multiple languages (English, Japanese, Chinese, and subsequently supports Vietnam). It can be used for efficient reasoning in CPU (ONNX) and NVIDIA GPU (PyTorch) environments. It has automatic voice splicing, voice mixing, and rich debugging and monitoring functions. It is suitable for developers and researchers to quickly integrate and deploy high-quality TTS services.
·Based on FastAPI and Docker, support rapid deployment and multi-architecture operation (CPU/GPU) ·Integrate Kokoro-82M TTS model to support English, Japanese, and Chinese (subsequent support for Vietnam) ·Compatible with OpenAI Speech API, supporting streaming Text To Speech and multiple output formats (mp3, wav, opus, flac, m4a, pcm) ·Support voice mixing (multiple sounds can be combined according to weights), and can export custom voice packets ·Provides advanced functions such as time-stamp subtitle generation for each word, word segmentation/phoneme level audio generation ·Built-in debugging interface to monitor system status, resource usage and inference session pool ·Support calls through REST APIs or OpenAI official libraries for easy integration into various applications ·Automatic clause splicing to optimize the long text processing experience
