Criteria for splitting on Labels

Audacity 3 Help Forum
1

I have a collection of stereo aiff files which I need to split into their constituent tracks; each aiff file has between 10 and 22 tracks.

The only way I have of telling where one track ends and the next begins is that I can see there is silence in between them. I have verified that the number of bursts (and thus the number of flat lines) corresponds to what I know to be the number of tracks for each aiff - using published Track Listings.

My questions, please, if someone patient enough to help me (new to Audacity)

  1. I can zoom in until I see what is obviously silence. I use Analyze > Measure RMS. Is this the best (only?) way to identify silence for the zone of each track which I can see to be silence?
  2. If not, what is the best metric, please?
  3. Is there then a way to load all mp3 tracks and have Audacity read all the silences to give me the thresholds I need?
  4. Assuming I can do that, when I come to Label the tracks using Analyze > Label sounds, which of the three possibilities (‘Peak’, ‘Average’, ‘RMS’) should I use to split?
  5. Can I automate this process with a Macro or script… there are 30+ mp3 files with a total of more than 500 individual tracks!

Many thanks in advance…!

2

If you are wanting to automate the process with a script, I think that ffmpeg can break up the audio into separate files based on silence. With the right parameters, probably with just one command, though you will probably have to practice with it.

If you are already familiar with using Nyquist code though, stick with the horse you know.

1 Like
3

Thanks, Rex!

I’m not familiar with either; though I bet I could learn.

As a first step I really need to know how to correctly/infallibly/consistently detect silence in Audacity across these many files, please.

4

Have you seen this: Label Sounds - Audacity Manual

1 Like
5

It looks like there is a program out there called mp3split that does this. I didn’t look long, so I don’t know if it is available for your operating system, and I don’t know if it has a GUI, or just uses command prompts.

1 Like
6

Many thanks, Rex. Yes in fact I looked at this before turning to Audacity. It seems to be over 10 years old; but I should look again, shouldn’t I?

Appreciated!

7

Many thanks, @billw58 ; Yes I used it to set things up in the first place. But I obviously got it wrong.

Is there any way that Audacity can scan all the files and then ‘report back’ on every RMS Threshold level and maximum and minimum durations of the silences?

If there were I could set the exporting according to Labels accurately.

Yesterday I prepared a spreadsheet of the 34 sets of files (500+ tracks in all) after splitting. Only a handful of split-detections were correct.

So my thought was to automate the whole process (with macros?) and keep running it with different parameters for silence detection until I cam somewhere close to have the correct number of split files in the 34 folders!