Aura 2 TTS is Deepgram's text-to-speech model built for production applications that demand natural-sounding speech with low latency. Deepgram, primarily known for its best-in-class speech recognition, developed Aura 2 to round out a full voice stack — letting developers use a single provider for both speech-to-text and TTS in voice pipelines.
The model emphasizes clean, professional-grade audio quality suitable for enterprise voice agents, customer service automation, and real-time voice interactions. It integrates directly with Deepgram's API infrastructure, making it straightforward to pair with their Nova STT models in bidirectional voice applications.
Key Features
Natural-sounding speech output with low time-to-first-byte
Multiple built-in voice personas suited for professional applications
Integrated with Deepgram's full voice API for unified STT+TTS workflows
Streaming audio delivery for real-time voice agents
Clear diction optimized for telephony and customer support contexts
Consistent quality across varied input sentence lengths
Ideal Use Cases
Automated voice responses in call center and IVR systems
Real-time voice agent pipelines paired with Deepgram STT
Notification and alert reading in enterprise monitoring tools
Podcast or audio content generation at scale
Accessibility narration for web and mobile applications
Example Prompts for Aura 2 TTS
Technical Specifications
| Provider | Deepgram |
| Category | Audio |
| Modality | Text -> Audio |
Frequently Asked Questions
Try Aura 2 TTS now
Start using Aura 2 TTS instantly — 100 free credits, no credit card required. Access 750+ AI models through one platform.