Dograh

Dograh provides an open-source, self-hostable platform for building and deploying robust voice agents. Key features include: • Bring your own speech-to-text, text-to-speech, and language models • Flexible deployment on your servers or their cloud • Model Context Protocol (MCP) for programmatic agent creation • Hybrid pre-recorded voice and speech generation for natural interactions • Real-time speech-to-speech option for ultra-low latency This platform allows developers to fully customize the voice agent stack, from inbound channels and telephony integration to the core language processing components. With the ability to swap out any node in the pipeline, users can plug in their preferred providers like Cartesia for STT, Deepgram for TTS, or Telnyx for SIP telephony, ensuring maximum control and flexibility. Dograh ships with a Model Context Protocol server, enabling programmatic voice agent development directly from IDEs. This allows development teams to build, modify, and deploy complex voice agents through code, streamlining the development process. The platform also features innovative hybrid voice generation, combining pre-recorded human clips with synthesized speech in the same cloned voice, significantly reducing latency and enhancing the naturalness of conversations. Ideal for engineering teams and businesses prioritizing data sovereignty, Dograh supports full self-hosting. This ensures lower latency and complete control over data, addressing critical needs for organizations that cannot use hosted voice agent vendors. It's a powerful alternative for those seeking an open-source, customizable voice interaction framework.