How to reduce sibilants from audio in Puppy? [solved]
[F96-CE 4]
I think the main Audio-Editing native program is gWaveEdit. I couldn't find this effect there. Does someone know how to do this in this program or another?
Discussion, talk and tips
https://forum.puppylinux.com/
[F96-CE 4]
I think the main Audio-Editing native program is gWaveEdit. I couldn't find this effect there. Does someone know how to do this in this program or another?
It looks like that can be done with audacity, google search for "De-Essing audacity linux how" : https://www.google.com/search?q=De-Essi ... +linux+how
Thanks, but Audacity is pretty big, and I don't need it for anything else (currently 1 file with speech, I want to listen to - and it's hardly bearable).
I think there was a good built-in program in Puppy, in the past. It could convert between files of many types. It may have had this filter - but don't remember; nor how it was called. And not sure if it was built-in. Anyway - hope for a small program.
I will probably work in the future with Kdenlive. Do you know if it has this ability?
Well, that's news to me! I've never heard of "de-essing" before.
Looks like a wee bit of research is in order, methinks....
Mike.
Sibilance, it's the audio term for overly loud "s" sounds: https://www.studiobinder.com/blog/what- ... efinition/
The most practical way to find tools for de-essing is to use an audio app that applies plugins, usually lv2 plugins, or vst plugins.
Unfortunately most of these audio apps are pretty big, and you have to download the plugins to the proper locations, usually /usr/lib/lv2
gwave-edit may be able to load more plugins than the built in effects, but I'd have to take a look a that to know for sure.
@mikewalsh - yeah, like geo_c said. I have a very good and sensitive Hearing; and there's a speech audio - that I can't listen to, because of the sharp S sound. I guess I'll have to just turn the volume very low, or listen with earplugs, or something.
If it's just for listening to a speech audio file, might it not be sufficient to take out the treble a bit, with (p)equalizer or other tone control?
...(p)equalizer or other tone control...
Where can I find and use them?
you should be able to attenuate higher frequency with use of ffmpeg. See here: https://superuser.com/questions/695843/ ... -on-ffmpeg
How do I use it, with a program??
ffmpeg is already installed and can use the terminal.
I tested with deadbeef (already has an equalizer builtin but disabled). I put the file takefive.flac as the source and output as take.flac.
on terminal (without the quotes):
"ffmpeg -i takefive.flac -af "equalizer=f=100:width_type=h:width=200:g=-50" take.flac"
played both files and source file sounds full, but take.flac with gain of -50 (negative), around the freq of 100, left with no bass, only mid-high tones in take.flac. In your case to attenuate the very high frequencies, u many want to change to f=1500 or even higher and it see if that helps.
OK, I made a few attempts. It definitely does something to the audio, but it's not optimal, and I have no idea what I'm doing. Can you explain the numbers, or know of a general explanation? For instance - which numbers will work only on the S sound (highest frequency?)?
MrAccident wrote: ↑Wed Sep 27, 2023 10:01 pmOK, I made a few attempts. It definitely does something to the audio, but it's not optimal, and I have no idea what I'm doing. Can you explain the numbers, or know of a general explanation? For instance - which numbers will work only on the S sound (highest frequency?)?
It depends on the particular audio file. If the "s" sound is raspy, I'd be focusing on the 700-2000 frequency range, if you're dealing with more of high hiss, almost whistly, then focus on 3000-6000. The width value, which I'm not exactly sure what the numbers are measuring, I'm pretty sure will affect frequencies fanning out from the center frequency you choose. Seems to be a parametric eq band. But I'd have to check on ffmpeg to see how it's configured.
Edit: Taking a look at this documentation, https://ffmpeg.org/ffmpeg-filters.html#equalizer , it looks like you can do a parametric "q" setting around a certain "f" frequency, what is known as a "notch-filter," but honestly, as one who does a lot of audio work, there will be a lot of trial and error using the command line version, because you have to basically guess at the frequency, q-factor, and gain cut, and then use your ears. And then reset, because if I understand correctly, after running one command, the next command will add another fiilter on top of that one, so first there would need to be some resetting to original output.
In DeaDBeef, menu VIEW > EQUALIZER
You can save the settings and load them later.
@fernan - worked. I just lowered gradually from 2.5 kHz and above (you can do that in one motion with the Mouse).
Thanks.