Speech-To-Speech-CPU ghcr.io/aurial/speech-to-speech-docker:latest https://ghcr.io/aurial/speech-to-speech-docker bridge sh false https://github.com/Aurial/Unraid-Community-Apps/issues https://github.com/huggingface/speech-to-speech https://github.com/Aurial/Unraid-Community-Apps#readme https://raw.githubusercontent.com/Aurial/Unraid-Community-Apps/main/templates/speech-to-speech-cpu.xml https://raw.githubusercontent.com/Aurial/Unraid-Community-Apps/main/s2s-icon.png AI:Tools:Utilities voice stt tts whisper realtime websocket conduit openai speech to speech assistant microphone transcription kokoro cpu Speech-To-Speech-CPU runs Hugging Face's open-source realtime voice-agent pipeline (Apache-2.0, 13k+ stars): voice activity detection, speech-to-text (faster-whisper), any OpenAI-compatible LLM as the brain, and Kokoro text-to-speech — served as one OpenAI Realtime WebSocket server. Point any Realtime-API client at it instead of OpenAI's cloud: Conduit's "local listening" (speech server), OpenAI Agents SDK clients, LiveKit, or its own browser demo. This template runs a CPU image built and published by Aurial (upstream ships CUDA-only and no registry image), so it needs no GPU and can run alongside GPU inference servers. It has no web page of its own: it is a backend that apps connect to. Configure the three LLM fields below to point at an existing OpenAI-compatible server (Strata, Radiance, Ollama, LM Studio, …); the key is stored masked, never on the command line. One voice session at a time by default; first start downloads the speech models (a few GB) into the cache folder, then it is warm and CPU-light. An OpenAI-compatible LLM server reachable over your network (Strata, Radiance, Ollama, etc.) — its URL and key go in the LLM fields below. The pipeline cannot start without it (it warms up the backend before opening the port). First start downloads STT/TTS models into the cache folder: allow 1–2 minutes before the port opens. Nothing else: pure CPU, no GPU flags, no driver plugin. 8765 /mnt/user/appdata/speech-to-speech faster-whisper kokoro cpu