Keywords not supported through SDK, or is it?

[tobyurff]

Which Deepgram product are you using?

Deepgram SDKs

Details

Hi there!

We're comparing STT providers at the moment and an essential bit for our use case is that transcription of known first and last name along with locations we know have a high change or happing during the stream is relatively accurate.

In the Node SDK, it's a bit unclear if that works, as there is keywords but it doesn't seem to do anything if specified:

deepgramStream = deepgram.transcription.live({
    smart_format: true,
    interim_results: true,
    language: "en-US",
    model: "nova",
    keywords: phrases, 
    keyword_boost: "legacy",
  });

Has anyone used this successfully for word hinting/custom vocabulary?

If you are making a request to the Deepgram API, what is the full Deepgram URL you are making a request to?

No response

If you are making a request to the Deepgram API and have a request ID, please paste it below:

No response

If possible, please attach your code or paste it into the text box.

deepgramStream = deepgram.transcription.live({
smart_format: true,
interim_results: true,
language: "en-US",
model: "nova",
keywords: phrases,
keyword_boost: "legacy",
});

If possible, please attach an example audio file to reproduce the issue.

No response

[jjmaldonis]

Hey @tobyurff, do you have spaces in your keywords? If so, try removing the spaces. You can also boost words more significantly by adding a :<number> after the keyword, such as Jason:3000. 3000 is extremely high, and numbers between 1-10 are more typical. You'll want to boost words higher if they have similar pronunciation to other words in your audio.

Also, is there a reason you're using keyword_boost: legacy? It sounds like you are most interested in boosting out-of-vocab keywords (names and places), so you will want to use the latest version of keyword boosting.