Diarization Inconsistencies with Nova-2 and Limited Language Support in Other Models

[rizz-official]

The diarization feature in Deepgram has the following issues:

Inconsistent Diarization with Nova-2:

Speaker detection is unreliable when using the nova-2 model with diarization enabled for supported languages.
Sometimes speakers are detected, but the diarization accuracy is low.
Diarization Limitations in Other Models:

Models like nova-phonecall and nova-medical do not support diarization for languages other than English.
Attempting diarization with non-English languages on these models results in a 400 Bad Request error.

[deepgram-community[bot]]

Thanks for asking your question. Please be sure to reply with as much detail as possible so the community can assist you efficiently.
_{Consider joining our Discord community for more opportunity to engage with your fellow Deepgram users. You can earn points which can be redeemed for cool stuff by being active in our communities!}

[deepgram-community[bot]]

Hey there! It looks like you haven't connected your GitHub account to your Deepgram account. You can do this at https://community.deepgram.com - being verified through this process will allow our team to help you in a much more streamlined fashion.

[deepgram-community[bot]]

It looks like we're missing some important information to help debug your issue. Would you mind providing us with the following details in a reply?

• The programming language you are working in (e.g. JavaScript, Python).
• The deepgram product you are using (e.g Speech to Text, Agent API)
• A request ID that triggered your error or issue.

[deepgram-community[bot]]

Yes, we've heard this a lot and understand the limits of Diarization. Our know our product team has been discussing improving it in 2025, but I don't have an ETA yet of when that improvement might be released.

Here are some suggestions to try:

1. Prepend audio from the primary speaker: For short audio files (under 3 minutes), prepend a 30-second clip of the primary speaker's voice before the full audio. This gives the diarization model a reference point to more reliably identify that speaker throughout the rest of the audio. <https://deepgram.gitbook.io/help-center/faq/improving-diarization-by-prepending-audio-from-the-primary-speaker|Help Center>
2. Use multichannel audio: When possible, use multichannel audio where each speaker is on their own channel. This can significantly improve diarization accuracy compared to mono audio with multiple speakers. <https://developers.deepgram.com/docs/diarization|Deepgram Docs>
3. Combine multichannel with diarization: For better performance, try using the multichannel feature together with diarization. This can help in cases where diarization alone might struggle. <https://github.com/orgs/deepgram/discussions/939|GitHub Discussion>
4. Use longer audio files: Diarization generally improves with longer audio files. If possible, provide longer audio samples to get better results.
5. Ensure good audio quality: Poor audio quality can significantly impact diarization performance. Avoid situations like holding a microphone up to a computer speaker, which can degrade the audio signal. <https://github.com/orgs/deepgram/discussions/283|GitHub Discussion>

This message was sent by John Vajda from Deepgram, via our community automation.