New Enhanced ASR Model Release
Deepgram is excited to announce our newest, most powerful ASR model, Enhanced model.
The new Enhanced model is based on our next generation End-to-End Deep Learning speech model architecture. Key features and benefits include:
- The new model has significantly higher accuracy and better word recognition. It has 19% higher accuracy relative compared to our previous model.
- It has increased effective vocabulary. It handles the long tail vocabulary significantly better.
Many of our customers have confirmed that it provides best in the industry accuracy with the highest speed right out of the box!
With the introduction of the new model, our customers can now select from three powerful solution tiers: our Base model, Enhanced model and our trained models.
You can start using the Enhanced Model in Console now by specifying model as ‘general-enhanced’. It is currently available for English.
Sample Curl code:
curl \ -X POST \ "https://api.deepgram.com/v1/listen?model=general-enhanced&punctuate=true" \ -H "Authorization:Token YOUR_SECRET" \ -H 'content-type: application/json' \ -d '{"url":"https://static.deepgram.com/examples/Bueller-Life-moves-pretty-fast.wav"}'
Read more about Deepgram Enhanced Model in the Deepgram Documentation.
Before using the new Model in production, we recommend testing it alongside your typical feature adds, like keywords. Model updates may change transcription outputs.
Deepgram support is always available to walk customers through updates, as well as resolve any issues that arise when upgrading.