AI transcribe for multiple tracks?

Ed,

There are 2 big issues here. 1) Transcribing as source media where the source is not audio track 1. 2) Editing using multitrack audio/i.e. multiple source transcripts.

See this staff post in the Beta forum:

https://community.adobe.com/t5/premiere-pro-beta-bugs/transcription-process-fails-if-no-audio-on-channel-1/idc-p/13742747#M595

I'm still experimenting with this. But since your audio track one is system sounds, it is no help for direct Source media transcription. (If one of your 2 dialogue tracks was track one, you could transcribe it.) But here's a method worth trying. Does it work for you?

Create one sequence from your clip. Let's call it Main Sequence. Unlink the audio; delete Audio track 2 and 3.

Now create a second sequence from the same clip, and call it Microphone. Be sure the timeline is active, switch to the Text Panel -> Transcript Tab. If you switch to the Source Monitor before getting to the Transcript tab, click back in the timeline or Program Monitor and then directly back to the Transcript tab. Click the 3 dots menu and pick Static transcript. In the settings, pick Audio track 2. Transcribe.

Repeat with another sequence called "Chat," and in the static transcribe pick Audio track 3.

Now go back to your Main sequence. Set the playhead to the beginning. In the Project Panel, right click on Micropohone sequence, and pick "Open in Source Monitor." Drat JUST the audio to Main sequence, audio track 2. Repeat with sequence Chat. You now have a Main sequence with 3 audio tracks, and 2 of them have Source media transcripts.

Now, Gotcha! You can only see the higher ranking transcript, meaning the one from Audio track 2, and not from 3. If there is a gap (not just a silence, but a break in the time), the Audio 3 transcript will show. If there are no gaps, to see Audio 3 transcript, either mute Audio 2 or drag Audio 2 to Audio 4.

I briefly extended this experiment to editing, since it is not very useful without an option. The Program Monitor/Timeline is Main sequence. In the transcript tab, we see the "source" transcript of nested sequence Microphone. The "Automatically set In/Out Points," is, I think, on by default. So highlight some text you want to remove. In/out points are marked in the Program Monitor and timeline. Your option is to Extract (ripple delete all tracks) or Lift (remove only the targeted track(s) and do not move on the timeline).

It would seem easiest to me to Lift only the track we see in the transcript (Audio track 2), so we target only it (only A2 is highlighted). Now mute A2 so A3 transcript shows. Select text there to delete and again Lift only track A3.

Bleeding edge. Still thinking about this. Sending this on to you to think about....

Stan