Batch select and delete a specific phrase

[Benjamin_Markus]

I'm a long time CC user, but relatively new to Audition. 

That said, we have a project with around 150 voice over recordings, listing off different amounts followed by the same two sentences.  The client has asked us to remove the second to last sentence from each recording and I'm wondering if there's a way to automate this in Audition.  It would save a lot of time if we didn't need to go through all tracks, manually select each similar waveform, delete them and export them all.

I noticed that there's a "Learn Sound Model" function in Effects > Noise Reduction / Restoration.  Could that sound model be used to learn the general shape of the sentence's waveform, and apply it to the other tracks?  If so, could we use that auto selection to batch delete it and export?  Or is there another method?

Thanks,

[SteveG_AudioMasters_]

There isn't really a way of doing this, I'm afraid - the 'learn sound model' is only really intended to remove short repeated sounds (not speech) and even that doesn't always work very well.

If the ends of these recordings is fixed and always identical, then a simple timing exercise that creates a favorite to remove a timed segment working back from the end rather than the start might do it - but if the sentence was respoken every time, then no automation system anywhere is going to achieve a workable result - you have to do it by ear.

Once you'd got into the swing of it, removing 150 sentences from individual files probably wouldn't take that long - probably less time than you'd spend experimenting with different methods and failing...

[Benjamin_Markus]

I was afraid of that, and yes you're right. It would take longer to experiment with other methods then to do it manually.

That said it should be possible to write some software that would do it. It would just take time to develop and I'm not sure there's a big enough demand for that specific functionality to make it a worthwhile endeavor.

 Thanks for the quick response though.

[SteveG_AudioMasters_]

That said it should be possible to write some software that would do it. It would just take time to develop and I'm not sure there's a big enough demand for that specific functionality to make it a worthwhile endeavor.

You might think that, but I'm afraid you'd be mistaken. Have a look at this BBC article, and you might begin to see how difficult this actually is...