Deepgram, a provider of speech recognition, natural language processing, and generative AI solutions, announced the public release of Aura, a text-to-speech (TTS) API that delivers human-like quality conversation. Aura is designed for developers who want to build real-time, conversational voice AI agents that can interact with customers, employees, and other users in a natural and engaging way. 

Aura can generate speech from any text input, including responses from LLMs like ChatGPT, in fractions of a second. This enables fluid and natural-sounding conversations with AI agents that can handle complex and dynamic scenarios. Aura offers a selection of diverse voices strongly suited for conversational use cases and preferences requiring the highest degrees of safety, security, speed, and scale.

Aura complements Deepgram’s Nova-2 speech-to-text API. With this release, Deepgram offers developers a complete voice AI platform, giving them the essential building blocks they need – from transcription to sentiment analysis to voice synthesis – to build high throughput, real-time AI agents of the future.