PolyAI Launches New Voice AI Model

Most voice AI systems convert speech to text before processing it. PolyAI changed this by building a model that understands raw audio input directly. This allows it to handle speech, timing, and actions all at once.

The model keeps text-to-speech separate so you can still control how the voice sounds. It runs as a standard request based system rather than a constant stream. This makes the technology easier to manage for developers.

Speed is the biggest advantage here. The company reports that the system responds in under 300 milliseconds in real world use. This helps make automated customer service calls feel much more natural.