speaches is an OpenAI API-compatible server supporting streaming
transcription, translation, and speech generation. Speach-to-Text is powered by
faster-whisper and for Text-to-Speech piper and Kokoro are used. This project
aims to be Ollama, but for TTS/STT models.