AssemblyAI announces general availability of its first text-to-speech model for voice agents

AssemblyAI describes its first text-to-speech model as generally available for live-call voice agents. It lists $15 per million input characters and quotes 90 ms to first audio in production. Separately, Artificial Analysis scored the model 81.7% and 16th on its Pronunciation Robustness benchmark. It reported generation at 162 characters per second, faster than five named models it compared.