Community

Keyterms Support in Deepgram's Nova Models

When working with Deepgram's API, it's important to leverage the API's capabilities to fit your specific needs, especially with tasks involving keyterms and advanced transcription. One commonly asked question pertains to the support for keyterms in different versions of Deepgram's nova models.

Keyterms in Nova Models

Keyterms are a critical feature for users who need to emphasize or track specific words or phrases during transcription. However, not all versions of Deepgram's models support this feature. As of the latest update, keyterms are supported exclusively in the Nova-3 model.

This implies:

Nova-2: Does not support keyterms. Users looking to utilize keyterms should transition to Nova-3 for full functionality or use Keywords.
Nova-3: Fully supports keyterms, allowing enhanced customization in transcription processes.

Considerations

Documentation: Always refer to the Deepgram keyterms documentation for the latest updates and corrections.
Model Comparison: If you're comparing transcription performance across models, be sure to note whether keyterms are active, as this can influence outputs significantly.

Conclusion

Always ensure you're using the correct model version that meets your transcription needs, especially concerning keyterms. For any issues or further clarifications, contacting support is recommended.

Reference Sources: