IMO it would also limit the possible use to FX that do nothing in the time domain. Release and attack times for compressors, decay time for reverbs are just two problem areas that I can quickly come up with at the chop up locations. And I’m sure there are more.
I tried today with dialogue audio of 1 hour, using auto split as specific time intervals, it didn’t really work, when I joined them back together, the start of each segment joined back up were jumpy. However, what did work was, I created 10 segments with region markers manually and timed the markers to where a person took a pause of silence, used auto split, put the split files into a batch for sonic treatment, then rejoined the files. it took 2 minutes to complete rather than 14min for the entire hour. I definitely think with these massive core counts there is something there to work within some way that could massively reduce render times, well, in the dialogue world anyway…
It would also require overlaps at the cuts of double the longest processing tail of any of the plugins in use
Exactly
Well, I am getting decent results using auto split, this morning a 1-hour single file render was going to take 28 min, I created 10 regions manually were the dialogue took a break speaking, used auto split to create 10 files, added 1 second top and tail, selected to create batch processor from a template, the processing took 5min which is a significant reduction. The only thing needed now is a solution to automatically compile the files back into 1 file again after the batch processing.
Well, I am getting decent results using auto split, this morning a 1-hour single file render was going to take 28 min, I created 10 regions manually were the dialogue took a break speaking, used auto split to create 10 files, added 1 second top and tail, selected to create batch processor from a template, the processing took 5min which is a significant reduction. The only thing needed now is a solution to automatically compile the files back into 1 file again after the batch processing.
Interesting to know. Maye I will do some experiments about this.
Philippe