Vocode is an open source platform that simplifies the creation of voice-based applications powered by large language models. It enables developers to build real-time streaming conversations with LLMs and deploy them across multiple channels including phone calls, Zoom meetings, and voice-based personal assistants. The library provides comprehensive abstractions and integrations that consolidate all necessary voice AI components into a single framework.

The platform supports both self-hosted deployments through Vocode Core and managed services via the Vocode API. Developers can spin up conversations using system audio, set up phone numbers with LLM-based agents for inbound and outbound calling, integrate with Zoom calls, and embed voice capabilities into Langchain agents. Vocode's architecture handles the complete voice pipeline from transcription through synthesis with built-in conversation functionality including endpointing and emotion tracking.

Vocode offers extensive integrations with industry-leading providers across the voice AI stack. For speech recognition, it connects with AssemblyAI, Deepgram, Gladia, Google Cloud, Microsoft Azure, RevAI, Whisper, and Whisper.cpp. Language model support includes OpenAI and Anthropic. Text-to-speech synthesis integrations span Rime.ai, Azure, Google Cloud, Play.ht, Eleven Labs, Cartesia, Coqui, gTTS, StreamElements, Bark, and AWS Polly. This breadth of integrations enables developers to select optimal providers for their specific use cases and requirements.