AssemblyAI has released AssemblyAI-TTS-1, a new text-to-speech API designed specifically for live phone call voice agents. Priced at $15 per 1M characters with no annual contract or concurrency tiers, the model closes the voice loop for developers already using AssemblyAI speech-to-text.
The hosted API operates on continuous audio representations rather than discrete tokens, trained on millions of hours of conversational data. It launches with a library of over 4,000 voices, instant voice cloning supporting six languages, and three integration interfaces including HTTP, WebSocket, and an OpenAI-compatible route. In self-reported tests, the model achieved a 47 ms median time to first byte and 93.7% entity accuracy.
Why it matters: This new API provides developers with an affordable, low-latency text-to-speech solution optimized for real-time conversational telephony systems.