Nova 2 transcription of multichannel, times are off

[cheike569]

Using nova-2 or nova-2-phonecall model and setting multichannel to true, the transcription times are off.

First word result for nova-2: Wrong

 {
    word: 'hello',
    start: 2.8799999,
    end: 3.12,
    confidence: 0.7920532,
    punctuated_word: 'Hello?'
  },

First word result for enhanced (old model): Correct

{
    word: 'hello',
    start: 0.03998,
    end: 0.53998,
    confidence: 0.7055664,
    punctuated_word: 'Hello?'
  },

Unfortunately, Github doesn't allow for audio files to be attached.

I'm using the JS SDK like this

 const {result, error} = await deepgram.listen.prerecorded.transcribeFile(
        fs.readFileSync(file),
        {
            model: "enhanced",
            punctuate: true,
            multichannel: true,
        }
    );

[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?

• A request ID that triggered your error or issue.

[jkroll-deepgram]

Hi @cheike569, are you able to share the Deepgram request ID, and/or a link to the audio (such as in Google Drive), so we can take a closer look?