A mono-repo to house the various supported Transport options to be used with the pipecat-client-web library. Currently, there are two transports: daily-transport and gemini-live-websocket-transport.
Pipecat Transports are intended to be used in conjunction with a Pipecat web client. Please refer to the full Pipecat client documentation here and an overview of the Transport API here
This Transport uses the Daily audio and video calling service to connect to a bot and stream media over a WebRTC connection. This Transport is the client-side counterpart to the Pipecat DailyTransport component.
Typical media flow using a DailyTransport:
┌────────────────────────────────────────────┐
│ │
┌───────────────────┐ │ Server ┌─────────┐ │
│ │ │ │Pipecat │ │
│ Client │ RTVI Messages │ │Pipeline │ │
│ │ & │ │ │ │
│ ┌──────────────┐ │ WebRTC Media │ ┌──────────────┐ media │ ┌─────┐ │ │
│ │DailyTransport│◄─┼────────────────┼─►│DailyTransport┼────────────┼─► STT │ │ │
│ └──────────────┘ │ │ └───────▲──────┘ in │ └──┬──┘ │ │
│ │ │ │ │ │ │ │
└───────────────────┘ │ │ │ ┌──▼──┐ │ │
│ │ │ │ LLM │ │ │
│ │ │ └──┬──┘ │ │
│ │ │ │ │ │
│ │ │ ┌──▼──┐ │ │
│ │ media │ │ TTS │ │ │
│ └───────────────────┼─┴─────┘ │ │
│ out └─────────┘ │
│ │
└────────────────────────────────────────────┘
This Transport extends the RealTimeWebSocketTransport and connects directly to Gemini over a WebSocket connection using the Multimodal Live API. This type of transport is great for testing different services out without the need to build a server component. Just be aware that it is insecure since you will need to have access to your Gemini API Key client-side so not probably something you want to use in your production app.
Media flow using a GeminiLiveWebSocketTransport:
Client Server
┌────────────────────────────────────┐
│ │
│ RTVIClient │ ┌──────────────┐
│ │ Media over │ │
│ ┌──────────────────────────────┐ │ WebSocket │ Gemini │
│ │ GeminiLiveWebSocketTransport │◄─┼────────────────┼─► Server │
│ └──────────────────────────────┘ │ │ │
│ │ └──────────────┘
└────────────────────────────────────┘
This Transport connects directly to OpenAI over a WebRTC connection using the RealTime API. This type of transport is great for testing different services out without the need to build a server component. Just be aware that it is insecure since you will need to have access to your OpenAI API Key client-side so not probably something you want to use in your production app. It does not implement the Ephemeral Token process.
Media flow using a OpenAIRealTimeWebRTCTransport:
Client Server
┌─────────────────────────────────────┐
│ │
│ RTVIClient │ ┌──────────────┐
│ │ Media over │ │
│ ┌───────────────────────────────┐ │ WebRTC │ OpenAI │
│ │ OpenAIRealTimeWebRTCTransport │◄─┼────────────────┼─► Server │
│ └───────────────────────────────┘ │ │ │
│ │ └──────────────┘
└─────────────────────────────────────┘
$ npm i
$ npm run buildBSD-2 Clause
We welcome contributions from the community! Whether you're fixing bugs, improving documentation, or adding new features, here's how you can help:
- Found a bug? Open an issue
- Have a feature idea? Start a discussion
- Want to contribute code? Check our CONTRIBUTING.md guide
- Documentation improvements? Docs PRs are always welcome
Before submitting a pull request, please check existing issues and PRs to avoid duplicates.
We aim to review all contributions promptly and provide constructive feedback to help get your changes merged.