TTS-1 from AssemblyAI – Frontier voice at $15/M chars (www.assemblyai.com)

AssemblyAI has launched TTS-1, a cutting-edge text-to-speech model now available to developers for just $15 per million characters. This model boasts ultra-low latency, delivering audio at just 90 ms, and allows for unlimited concurrent usage without a contractual commitment. This affordability, coupled with its nuanced human-like speech, addresses two significant barriers to the widespread adoption of voice agents: the need for natural-sounding interaction and cost-effective scaling for businesses shifting from traditional phone systems to automated services. TTS-1 is built on a foundation of millions of hours of conversational audio, leveraging advanced techniques that enhance audio realism by operating directly on continuous rather than discrete audio representations. With an impressive 99% word accuracy and quick response times, it outpaces competitors, making it the fastest in terms of audio delivery. The product not only supports over 4,000 voices and various accents but also offers features such as instant voice cloning for personalized user experiences. As voice agent technology evolves, TTS-1’s performance metrics and pricing structure position it as a transformative option for commercial applications in AI-driven communication.