Text-Based-Editing Doesn't work with multiple audio channels

I am getting this error even though Ihave audio in track 1. I admit that I am confused about the terminology being used - in the sequence I see two stereo audio tracks A1 (Ch2, Ch 3) and A2, but when I look at mondify audio it refers to them as Clip 1 and Clip 2 - Each with L-1, R-1, 1, 2. Not sure how to make sense of this.

I personally think that the easiest and most obvious solution to this issue would be for them to allow you to select which tracks you want to transcribe with a checkbox. Any tracks selected for transcription could be merged into a single mono audio track for transcription. Yes, I can think of some edge cases where this might not generate optimal results (a lav that is temporarily off camera, for instance) but I think it's a far superior approach versus what is happening right now.

I ran into this issue not too long ago and doing the Generate Transcript function seemed to help in this particular situation.

@kit.kohler,

Here's a feature request in the regular forum:

https://community.adobe.com/t5/premiere-pro-ideas/audio-track-selection-for-source-media-transcription/idi-p/13828019

I have thought that this could be more difficult than it looks: as rolled out, they only have to track one audio track per clip. And there can only be one transcription per clip. Depending on how this is implemented, you could have numerous transcripts per clip, with numerous clips per sequence. My extracted audio workaround gets a little crazy! Users are having some trouble just figuring out source transcripts vs sequence static transcripts.

But that is not an argument against selecting which audio you want to transcribe. I just think it may have some difficult choices.

@nbechere, I'm assuming there when you guys are ready, there will be a Beta conversation on this similar to the earlier text-based elements?

Stan

Also upvoting this. I was totally bewildered about why my transcriptions weren't working until I dug into the forums here. You've put all the development into this, being able to select which track to transcribe cannot possibly be that difficult to accomplish.

I appreciate staff from Adobe reading and responding to issues like this, it shows willing and I'm grateful. For the information of others who want to try this workaround as I did - it works in theory but can't maintain timecode when a nest in a new sequence - it just reverts back to TC starting 00. I have to join with many others in complaining that this feature is not ready for market as professionals require it, and to claim otherwise is not honest. It could have easily been predicted that professionals would need multi-channel audio choices for transcriptions. The lack of it causes huge problems for timecode continuity across multi-channel sequences with cuts. Fixing this can't be honestly described as a "future enhancement", it's a basic requirement that we paid money for without being informed wouldn't come with it.

@nbechere Nico,

That is a great tip. I can think of other ways that this would be helpful.

I took the warning for a static transcript (won't update with changes in the sequence) as meaning they were different creatures. The transcription is static as long as you are using it as a sequence. But as soon as you use it as a nest in a sequence - or open it in the Source Monitor for 3-point editing as a source file, all the text-based editing options will be there.

Stan

Thanks for that feedback. Yes, this is on our radar as a future enhancement. The main issue I see is that this would not work with background transcription enabled but it would surely work with manual transcriptions. 

What you can do right now as a workaround (without re-encoding) is to wrap your source in a sequence, transcribe that sequence statically (via the text panel overflow menu -> Generate static transcript) and select mixdown or the relevant tracks there. You can then use the transcribed sequence as source footage for your timeline edit. 

Upvoted. There have been other reports that this is a problem.

The option to pick any one track would help. I suspect someone will have a use case where mix would be desireable.

The problem case would be where you have the audio for 2 interviewees and really need the Speaker identification and/or detection.

You describe the workaround - reencoding.

This is not a bug as such, but there is no Ideas subforum in Beta. But staff treat every discussion as Ideas!

Stan