Deepgram Nova-3 released on Vapi

Deepgram's Nova-3 model, now integrated into VAPI, significantly enhances speech recognition capabilities with a notable 54.3% reduction in word error rate for streaming, setting a new accuracy standard. It is the first in the industry to provide real-time multilingual transcription, demonstrating superior performance across seven tested languages. The model also offers revolutionary self-serve customization for immediate vocabulary adaptation without the need for model retraining, making it ideal for complex enterprise applications and challenging audio conditions. Integration is straightforward, requiring only a model name change from Nova-2 to Nova-3 in configurations, and is currently available through Deepgram's hosted endpoint. For additional information, users are encouraged to consult Deepgram's launch post or contact their support team.