Speaches: Local API server for TTS/STT models using faster-whisper