deepgram / deepgram-js-captions Goto Github PK

What is the current behavior?

If speaker diarization is used, the voice tags <v Speaker> are not closed. The closing tag is missing.

Steps to reproduce

Convert deepgram output with different speakers to VTT.

Expected behavior

<v Speaker>Hello!</v>

What is the current behavior?

I am seeing a TypeError when using the srt parser from the @deepgram/captions library:

TypeError: Cannot use 'in' operator to search for 'speaker' in undefined
    at DeepgramConverter.getLines (.../@deepgram/captions/src/converters/DeepgramConverter.ts:22:32)
    at srt (.../@deepgram/captions/src/index.ts:54:20)

Steps to reproduce

This error is occurring when using the deepgram nova-2 model via the REST api:

      method: 'POST',
      url: 'https://api.deepgram.com/v1/listen',
      params: {
        model: 'nova-2',
        punctuate: true,
        smart_format: true,
        paragraphs: true
      }

Expected behavior

The srt parser works on the nova-2 output without any errors

Please tell us about your environment

• node v18.19.1