# Voice agents and realtime speech

What a spoken agent is built from: frameworks that manage a live conversation, hosted voice-agent platforms, speech-to-speech model APIs, and the speech recognition and synthesis models and services underneath.

44 tools, in alphabetical order. Nothing is ranked by popularity.

- [AgentPhone](https://modelmoot.com/tools/agentphone.md): A service that gives AI agents their own phone numbers for calls and text messages through an API. (Relay Innovations; not open source, hosted service)
- [Agora Conversational AI Engine](https://modelmoot.com/tools/agora-conversational-ai.md): A hosted engine from Agora that joins realtime audio channels with a speech and language model pipeline for voice agents. (Agora; not open source, hosted service)
- [AssemblyAI](https://modelmoot.com/tools/assemblyai.md): A speech API for transcription and streaming speech recognition, with a voice agent offering. (AssemblyAI; not open source, hosted service)
- [Bland AI](https://modelmoot.com/tools/bland-ai.md): A hosted platform for building AI phone agents that place and answer calls through an API. (Bland AI; not open source, hosted service)
- [Bolna](https://modelmoot.com/tools/bolna.md): An open-source framework and platform for building voice agents that place and take phone calls. (Bolna; open source, hosted service)
- [Cartesia](https://modelmoot.com/tools/cartesia.md): A hosted speech platform with low-latency text-to-speech and speech-to-text APIs aimed at voice agents. (Cartesia; not open source, hosted service)
- [Chatterbox](https://modelmoot.com/tools/chatterbox.md): An open text-to-speech model from Resemble AI with voice cloning and expressiveness controls. (Resemble AI; open source)
- [Cognigy](https://modelmoot.com/tools/cognigy.md): A conversational AI platform for customer service in contact centers, offered as NiCE Cognigy with voice and chat channels. (Cognigy; not open source, hosted service)
- [Coval](https://modelmoot.com/tools/coval.md): A hosted platform for testing and evaluating voice AI agents before launch and monitoring them in production. (Coval; not open source, hosted service)
- [Daily](https://modelmoot.com/tools/daily.md): A realtime video and audio infrastructure service with SDKs for building voice agents, from the makers of Pipecat. (Daily; not open source, hosted service)
- [Decagon](https://modelmoot.com/tools/decagon.md): A platform for companies to deploy customer support agents on chat, email and voice. (Decagon; not open source, hosted service)
- [Deepgram](https://modelmoot.com/tools/deepgram.md): A speech platform with APIs for transcription, speech synthesis and a voice agent API. (Deepgram; not open source, hosted service)
- [ElevenLabs Agents](https://modelmoot.com/tools/elevenlabs-agents.md): A hosted platform for building conversational voice agents on ElevenLabs speech models, with telephony and tool support. (ElevenLabs; not open source, hosted service)
- [faster-whisper](https://modelmoot.com/tools/faster-whisper.md): A reimplementation of Whisper using CTranslate2 that runs transcription faster and with less memory. (SYSTRAN; open source)
- [Gemini Live API](https://modelmoot.com/tools/gemini-live-api.md): A Google API for low-latency voice and video conversations with Gemini models over WebSockets. (Google; not open source, hosted service)
- [Hamming](https://modelmoot.com/tools/hamming.md): A hosted platform for testing, red-teaming and monitoring voice and chat AI agents. (Hamming; not open source, hosted service)
- [Hume EVI](https://modelmoot.com/tools/hume-evi.md): A realtime voice interface API from Hume that adds emotion-aware speech and expression analysis to conversations. (Hume AI; not open source, hosted service)
- [Kokoro](https://modelmoot.com/tools/kokoro.md): A small open text-to-speech model with a Python inference package, released with Apache licensed weights. (hexgrad; open source)
- [Kore.ai](https://modelmoot.com/tools/kore-ai.md): An enterprise platform for building and managing agents, search and conversational applications for employees and customers. (Kore.ai; not open source, hosted service)
- [LiveKit Agents](https://modelmoot.com/tools/livekit-agents.md): A framework for building realtime voice and video AI agents that join LiveKit rooms, in Python and Node.js. (LiveKit; open source, hosted service)
- [Moshi](https://modelmoot.com/tools/moshi.md): A speech-text foundation model and full-duplex spoken dialogue framework that listens and speaks at the same time. (Kyutai; open source)
- [Okareo](https://modelmoot.com/tools/okareo.md): A platform for testing, evaluating and monitoring voice, text and headless AI agents with synthetic users. (Okareo; not open source, hosted service)
- [OpenAI Agents SDK for Python](https://modelmoot.com/tools/openai-agents-python.md): OpenAI's Python library for agents with tools, handoffs between agents, guardrails and tracing. (OpenAI; open source)
- [OpenAI Agents SDK for TypeScript](https://modelmoot.com/tools/openai-agents-js.md): OpenAI's TypeScript library for agents with tools, handoffs, guardrails, tracing and realtime voice. (OpenAI; open source)
- [OpenAI Realtime API](https://modelmoot.com/tools/openai-realtime-api.md): An API for low-latency speech-to-speech conversations with OpenAI models over WebRTC, WebSocket and SIP. (OpenAI; not open source, hosted service)
- [Pipecat](https://modelmoot.com/tools/pipecat.md): An open-source Python framework for building voice agents and multimodal realtime applications as pipelines of services. (Daily and the Pipecat community; open source, hosted service)
- [Piper](https://modelmoot.com/tools/piper.md): A fast local neural text-to-speech engine that runs on small devices, now maintained by the Open Home Foundation. (Open Home Foundation; open source)
- [Retell AI](https://modelmoot.com/tools/retell-ai.md): A hosted platform for building and deploying voice agents for phone calls, with an API, SDKs and a dashboard. (Retell AI; not open source, hosted service)
- [Rime](https://modelmoot.com/tools/rime.md): A hosted text-to-speech service with models designed for realtime voice agents and phone calls. (Rime; not open source, hosted service)
- [Sesame CSM](https://modelmoot.com/tools/sesame-csm.md): An open conversational speech generation model that produces speech from text and audio context. (Sesame; open source)
- [Sierra](https://modelmoot.com/tools/sierra.md): A platform for companies to build and run customer service agents for chat, voice and email. (Sierra; not open source, hosted service)
- [Silero VAD](https://modelmoot.com/tools/silero-vad.md): A small voice activity detection model that finds speech in audio, often used for turn-taking in voice agents. (Silero; open source)
- [Speaches](https://modelmoot.com/tools/speaches.md): An OpenAI-API-compatible server for speech-to-text and text-to-speech that runs locally. (Speaches contributors; open source)
- [Speechmatics](https://modelmoot.com/tools/speechmatics.md): A speech recognition and voice agent platform with realtime transcription, speaker diarisation and many languages. (Speechmatics; not open source, hosted service)
- [Synthflow](https://modelmoot.com/tools/synthflow.md): A hosted no-code and API platform for building voice agents that handle phone calls. (Synthflow; not open source, hosted service)
- [Telnyx](https://modelmoot.com/tools/telnyx.md): A telecom platform on its own carrier network with voice AI agents, SMS, phone numbers and a hosted MCP server. (Telnyx; not open source, hosted service)
- [TEN Framework](https://modelmoot.com/tools/ten-framework.md): An open framework for building conversational voice agents from extensions for models, speech and transport. (TEN-framework contributors; not open source)
- [Twilio](https://modelmoot.com/tools/twilio.md): A communications platform with APIs for SMS, voice, WhatsApp, email and other channels, with an MCP server for its APIs. (Twilio; not open source, hosted service)
- [Twilio ConversationRelay](https://modelmoot.com/tools/twilio-conversationrelay.md): A Twilio feature that connects phone calls to a language model over WebSocket with speech recognition and synthesis handled by Twilio. (Twilio; not open source, hosted service)
- [Ultravox](https://modelmoot.com/tools/ultravox.md): A speech-native language model and hosted API for realtime voice agents, with open model weights. (Ultravox; open source, hosted service)
- [Vapi](https://modelmoot.com/tools/vapi.md): A hosted platform and API for building voice agents that make and take phone and web calls. (Vapi; not open source, hosted service)
- [Vocode](https://modelmoot.com/tools/vocode.md): An open-source Python library for building voice-based LLM agents with modular speech and telephony components. (Vocode; open source)
- [Whisper](https://modelmoot.com/tools/whisper.md): An open-source speech recognition model family from OpenAI that transcribes and translates audio. (OpenAI; open source)
- [whisper.cpp](https://modelmoot.com/tools/whisper-cpp.md): A C and C++ port of Whisper for running speech recognition on laptops, phones and servers without Python. (ggml-org; open source)

HTML: https://modelmoot.com/tools/c/voice. All categories: https://modelmoot.com/tools.md
