Latest / Kana & Mari's SoundRepos (English)

zaigie/FunSpeech | private ASR and TTS API gateway | GitHub
FunSpeech is a locally deployable, private speech platform built as a microservice architecture for automatic speech recognition and text-to-speech. Its CPU gateway provides Aliyun and OpenAI compatible HTTP APIs plus WebSocket streaming, while isolated GPU services support FunASR, Dolphin, Qwen3-ASR, CosyVoice, Qwen3-TTS, and vLLM-Omni backends with voice cloning and scalable Docker Compose deployment. Related: zaigie/FunSpeech / private ASR and TTS API gateway / GitHub
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.