WaveFeedback
All ideas
CompleteCharles Ballentine · 10mo ago

Globally edit speakers in transcript

I practiced speaker isolation and identification by recording a podcast. The app did an excellent job of accurate speaker separation even though they occasionally talked over each other. Impressive! The GUEST was correctly identified and labeled in the transcript. However, the HOST was incorrectly identified as being the name of the guest introduced for the next segment, which I did not record. In many cases, the transcript will be much more useful to me than the summary, so I wanted to figure out a way to edit the transcript to reflect the speakers accurately. The app lets me edit each speaker's line individually but not globally. i suppose. I could export it to a word processor and do it in there, but it would be nice if I could globally edit it in the app. Not an issue, but a comment: While the app was processing the translation, the word "Recording..." was still visible in the upper left-hand corner, leading me to believe that it was still RECORDING the conversation rather than PROCESSING it. I started frantically looking for ways to stop the recording when the translation finally appeared.

3 comments

  • Ryan Goble10mo ago

    I wish I could upvote this more than once.

  • Mac Connolly10mo ago

    Yes me too! Some of the other apps like Otter.ai and Zoom will let you tags once, and re-analyze across the entire transcript! It’s even learns for next call. I have the same 50 or so international guys on my team - its butchers their names, calls them the same name by ethnicity, and can’t hardly make out what they’re saying. I get it? Accents are hard. But doesn’t seem like it learns at all.

  • David Morris10mo ago

    There can never be enough upvotes on this. The bottom line is that it needs to be able to accept user corrections and LEARN from them to reanalyze the output. This seems to be a "one-and-done" app, and the only solution is to edit everything manually. It would help if Wave.AI could recognize "favorite" speakers. I meet with the same people regularly and should be able to teach who those voices are, so I don't need to make the same changes over and over again.