Speech-to-speech
Speech-to-speech
Realtime models that take audio in and give audio out, in one service.
A speech-to-speech service replaces the STT, LLM and TTS trio with one realtime model: audio in, audio and text out, with function calls.
| Provider | Classes | Package | Status |
|---|---|---|---|
| OpenAI Realtime | OpenAIRealtimeLLMService | @pipedog/openai | Ported |
| OpenAI Live | OpenAILiveLLMService | @pipedog/openai | Ported |
| Gemini Live | GeminiLiveLLMService | @pipedog/google | Ported |
| Gemini Live on Vertex AI | GeminiLiveVertexLLMService | @pipedog/google | Ported |
| Inworld Realtime | InworldRealtimeLLMService | @pipedog/inworld | Ported |
| Ultravox | UltravoxRealtimeLLMService | @pipedog/ultravox | Ported |
| xAI Grok Realtime | GrokRealtimeLLMService | @pipedog/xai | Ported |
| AWS Nova Sonic | AWSNovaSonicLLMService | @pipedog/aws | Ported |
| Azure Voice Live | AzureVoiceLiveLLMService | @pipedog/azure | Ported |
| Azure OpenAI Realtime | AzureRealtimeLLMService | @pipedog/azure | Ported |
Pipecat's speech-to-speech reference lists every upstream provider. Those not in this table are on Other upstream providers.
pipedog is an unofficial community TypeScript port of Pipecat. It is not affiliated with or endorsed by Daily or the Pipecat project.
Portions adapted from the Pipecat documentation, Copyright (c) 2024–2025 Daily, BSD-2-Clause (see apps/docs/LICENSE-pipecat-docs).