Finding tracks

Programming and Development Macros and Scripting
1

Screenshot from 2019-10-01 14-31-48.png
Hi,
I’m working with the David W Niven class jazz tape recordings. (https://archive.org/details/davidwnivenjazz). These tapes are each about 50 minutes long and consist of tapes made from old 78s and LPs interespersed with commentary and liner notes. They are a wonderful resource; however, they are a bit intimidating because of their length.

I want to split them into mp3s each containing a single track and associated commentary. I hope to add them to the repository. I’ve done a few by hand but as there are over 600 and each has two sides, I would like to automate the process.

The commentary is on one track, music on both - even when the original recording was mono, In the example shown, (https://archive.org/download/Billie_Holiday_Tape_2_1936-1937/Billie_Holiday_Tape_2_1936-1937_Side_A.mp3) I would like breaks at just under 4 minutes, 7 1/2 minutes, 11 minutes, … I proceed by opening the file, spliting to mono, selecting the track with the quiet sections (this is the music track) and using find silence / sound to start. The older recordings are very noisy and neither finding silence nor sound is robust.

I believe that with the ability to script and control parameters for finding sound&silence, I could solve the problem but I don’t know whether this is possible in Audacity. I’m really looking for ideas about how to proceed. Any suggestions/pointers would be greatly appreciated.

Thanks.
–Len

2

There’s a “trick” that can help sound / silence finder to work better with noisy recordings.

The noise in old recordings is often mostly high frequencies (clicks and crackles). These can be reduced considerably by applying a “Low pass filter” (see: https://manual.audacityteam.org/man/low_pass_filter.html). A low pass filter will also make the recording sound muffled, so you won’t want to apply it to the actual audio, but you can apply it to a copy of the audio.
Another common noise is “rumble”. This can be reduced with a “High Pass Filter” set at around 100 Hz.

So here are some steps:

  1. Import the audio track
  2. With the track selected, “Edit menu > Duplicate” (Ctrl + D)
  3. Ensure that only the duplicate track is selected, and apply the Low Pass Filter (try 200 Hz, 48 dB per octave)
  4. Apply the High Pass Filter (100 Hz)
  5. Apply the “Normalize” effect to bring the overall level of the duplicate track back up.
  6. Apply Sound / Silence finder to the duplicate track
  7. Delete the duplicate track by clicking the [X] in the top left corner of the track.

Tip: As you want to split on gaps in the music, after step 2, “split the track to mono” (https://manual.audacityteam.org/man/audio_track_dropdown_menu.html#stereo) and delete the voice channel. You can then use Silence finder to find long silences (say longer than 5 seconds).

With the track that you provided, the above worked pretty well for me, though as you can see, it missed a couple of track marks, and gave me one duplicate label.
Tracks000.png