Mastering Transcription with Speaker Diarization Tips (Full Transcript)

Speaker 1: Oftentimes, when you're working with audiophiles, you're going to want a transcription. But what happens when you have multiple speakers, for example, in a meeting with multiple attendees, or if you're looking at a podcast with a couple of guests on the podcast where you want to parcel out each person's individual contribution? That's a particular type of transcription called speaker diarization, and I'm going to show you how I do it on my end, leveraging a tool called Assembly AI. I'm here in a notebook, and I'm leveraging their API. You can input your API key to follow along with this particular example, and I'm grabbing an audiophile. I oftentimes will upload my audiophile directly here to my notebook, and then the config for the API call, I'm saying speaker labels true. That's important so you get the appropriate transcript that I'm going to show you here. When you run that, what you're going to get back is you're going to get a transcription that's going to be parceled out by speaker. You have speaker A, speaker B, speaker A, speaker B. That's pretty helpful. The next thing that I like to do is I like to grab this, go over to ChatGPT, and just have ChatGPT reformat this in the way that I want it formatted. It might say, hey, reformat this, reformat this, format this, and I'm going to just point that to this. Speaker A is Manny, and then speaker B is Jenny. Then I'll just copy, paste that there, and ChatGPT will go ahead and reformat this. I'll say, hey, Manny says this, Jenny says this, Manny says this. Again, this is super helpful for me when I transcribe podcasts that I'm a big fan of, and I want to see what the interviewer is saying and what the guest is saying. I hope that's a helpful tip for you to add to your toolbox. If you have any questions, feel free to drop it below. Cheers.