Rapida Voice AI is an open-source, end-to-end voice AI orchestration platform for building real-time conversational voice agents. It handles the full call stack, from audio streaming and speech recognition to telephony and observability. The project has about 748 stars on GitHub and is under active development, with updates in October 2026. Note that this entry covers Rapida’s open-source voice platform, a distinct project from a similarly named product already listed on Tooloso.
Written in Go and built on gRPC for low-latency bidirectional audio, Rapida provides both a platform and a framework for teams that need ownership of their voice stack. It combines speech-to-text, text-to-speech, voice activity detection, agent state management and multi-channel integrations such as SIP telephony in one system. Operators can bring their own models from providers like OpenAI or Anthropic, or use open-source models, avoiding vendor lock-in. Agencies use it for white-label client deployments, while enterprises run it as internal voice infrastructure, either managed or fully self-hosted.
Rapida is freemium: the platform is open source under GPL-2.0 with additional conditions, free to self-host. A commercial license is available for enterprise use, covering branding removal and closed-source deployments, and managed hosting is offered through rapida.ai.
Rapida suits agencies building voice agents for multiple clients and enterprises that need scalable, controllable voice infrastructure for support lines, receptionists or outbound calling. It is a strong fit when data ownership and deployment flexibility matter more than a plug-and-play hosted tool.