#172 logarithmic fade only spans across half the fade-out-length

open
None
5
2011-10-14
2011-10-13
Anaminus
No

Whenever I apply a logarithmic fade, half of the fade ends up being complete silence. This doesn't occur with other fade types.
i.e. If I wanted to have 10 seconds of correct fade-out at the end of the file, I'd have to apply 20 seconds of fade-out during the last 10 seconds, then trim off the other 10 seconds of appended silence.

Discussion

  • Ulrich Klauer

    Ulrich Klauer - 2011-10-13

    This works for me. However, after one half of the fading process, the signal is indeed very quiet already: -50 dB, so if it started at perhaps -12 dBFS, it is at -62 dBFS ten seconds in a twenty-seconds fade. This means it is below the resolution of 8-bit files, and possibly below a person's hearing threshold depending on type and volume setting of loudspeakers/headphones, ambient noise, etc. Is it possible that is what you noticed?

    I did a few experiments like this
    sox -n -d synth 20 sin 440 vol -12dB fade l 0 20 20 trim 0:10 1 stats
    varying the start time (10) of the trim effect. On my loudspeakers, I can still hear something at 0:11 seconds, but not at 0:12 anymore. Still, stats shows a signal is present at -72 dBFS, as it should be.

    (Technically, logarithmic fade never completely fades out. At the end of the fade, we're at -100 dB, still well within the dynamic range of 24-bit audio.)

     
  • Anaminus

    Anaminus - 2011-10-14

    I'm not too keen on the science of it, but I understand that that's probably the case.

    Other types did sound as they should have, but logarithmic was the closest to what I was listening for. Perhaps more fade types, or even the ability to customize the fade, would make a good feature?

     
  • Ulrich Klauer

    Ulrich Klauer - 2011-10-14
    • assigned_to: nobody --> uklauer
     
  • Ulrich Klauer

    Ulrich Klauer - 2011-10-14

    OK, I'll make this a feature request. I'm actually already working on a more flexible approach to fading, including fading not to/from zero. But this will take a while.

     
  • Anaminus

    Anaminus - 2011-10-15

    Awesome! Thanks for the response. I'm looking forward to the future developments.

     

Log in to post a comment.