a stereo (music) to 5.1 with Moitah's CenterCut GUI and Bidule

[Yuri Scherbakov]

Hi!
I've studied and tried out every possible method of a stereo to 5.1ch music (not movie) material conversion described in this forum (http://forum.doom9.org/showthread.php?t=83752&highlight=stereo+conversion) and wasn't quite happy with the results. Either the stereo image is jumbled up (V.I Stereo to 5.1 Converter VST Plugin Suite, the SRS plug-in, and all Bidule Ambisonics built-ups at http://www.dtsac3.com/cgi-bin/yabb/YaBB.pl?board=3001;action=messageindex;start=20 or overflanged and distorted output produced by the very time consuming Adobe Audition 2's filter. When I tried Moitah's stand-alone http://www.moitah.net/download/latest/Center_Cut_GUI.zip it did the sleek job of a music stereo to center and surround channels separation. The algo is here http://www.virtualdub.org/blog/pivot/entry.php?id=102.
The discussion's here http://forums.virtualdub.org/index.php?&act=ST&f=21&t=12627&st=0
Many thanks to Moitah (http://www.moitah.net/).
Now that I have the unchanged stereo, a mono of the center channel, and a stereo of flawless surrounds, I can feed them into Bidule for the second stage of processing which gives the LEF channel (stereo expander (-100) - Waves MaxxBass - Waves RBass), dithering, the 90 degree phase-shift of the surrounds, the 24 bit conversion (I end up with DVD-Audio format), and the required number-of-channels output - either three 2-channel files for my purpose, or one 6-channel file for somebody else's for AC3 or DTS CDs.
Finally, I load the three stereos into WaveLab 5's DVD-Audio montage - 2 of them (2 albums, 24-bit resolution) per one DVD disk and burn.

[kikatu]

Hello Yuri, I’m glad this method worked best for you; depending on the content or simply our personal preference some guidelines might work better than others. :goodpost:

[raquete]

cool center only(hard) and L&R-C(surrounds)
L&R+C give the stereo source again without looses,i never saw this before.:eek:

it's a treasure Yuri.
thanks for this post here.

[Yuri Scherbakov]

Thank you! But let all our thanks go to Moitah! He's done outstanding work!

[nivedhya]

Hi!
I've studied and tried out every possible method of a stereo to 5.1ch music (not movie) material conversion described in this forum (http://forum.doom9.org/showthread.php?t=83752&highlight=stereo+conversion) and wasn't quite happy with the results. Either the stereo image is jumbled up (V.I Stereo to 5.1 Converter VST Plugin Suite, the SRS plug-in, and all Bidule Ambisonics built-ups at http://www.dtsac3.com/cgi-bin/yabb/YaBB.pl?board=3001;action=messageindex;start=20 or overflanged and distorted output produced by the very time consuming Adobe Audition 2's filter. When I tried Moitah's stand-alone http://www.moitah.net/download/latest/Center_Cut_GUI.zip it did the sleek job of a music stereo to center and surround channels separation. The algo is here http://www.virtualdub.org/blog/pivot/entry.php?id=102.
The discussion's here http://forums.virtualdub.org/index.php?&act=ST&f=21&t=12627&st=0
Many thanks to Moitah (http://www.moitah.net/).
Now that I have the unchanged stereo, a mono of the center channel, and a stereo of flawless surrounds, I can feed them into Bidule for the second stage of processing which gives the LEF channel (stereo expander (-100) - Waves MaxxBass - Waves RBass), dithering, the 90 degree phase-shift of the surrounds, the 24 bit conversion (I end up with DVD-Audio format), and the required number-of-channels output - either three 2-channel files for my purpose, or one 6-channel file for somebody else's for AC3 or DTS CDs.
Finally, I load the three stereos into WaveLab 5's DVD-Audio montage - 2 of them (2 albums, 24-bit resolution) per one DVD disk and burn.



I'm unable to get that center_cut_GUI.Where can I get?

[nivedhya]

Hi!
I've studied and tried out every possible method of a stereo to 5.1ch music (not movie) material conversion described in this forum (http://forum.doom9.org/showthread.php?t=83752&highlight=stereo+conversion) and wasn't quite happy with the results. Either the stereo image is jumbled up (V.I Stereo to 5.1 Converter VST Plugin Suite, the SRS plug-in, and all Bidule Ambisonics built-ups at http://www.dtsac3.com/cgi-bin/yabb/YaBB.pl?board=3001;action=messageindex;start=20 or overflanged and distorted output produced by the very time consuming Adobe Audition 2's filter. When I tried Moitah's stand-alone http://www.moitah.net/download/latest/Center_Cut_GUI.zip it did the sleek job of a music stereo to center and surround channels separation. The algo is here http://www.virtualdub.org/blog/pivot/entry.php?id=102.
The discussion's here http://forums.virtualdub.org/index.php?&act=ST&f=21&t=12627&st=0
Many thanks to Moitah (http://www.moitah.net/).
Now that I have the unchanged stereo, a mono of the center channel, and a stereo of flawless surrounds, I can feed them into Bidule for the second stage of processing which gives the LEF channel (stereo expander (-100) - Waves MaxxBass - Waves RBass), dithering, the 90 degree phase-shift of the surrounds, the 24 bit conversion (I end up with DVD-Audio format), and the required number-of-channels output - either three 2-channel files for my purpose, or one 6-channel file for somebody else's for AC3 or DTS CDs.
Finally, I load the three stereos into WaveLab 5's DVD-Audio montage - 2 of them (2 albums, 24-bit resolution) per one DVD disk and burn.


center_cut_GUI & dsp centercut same or different?

I'm interested in your above conversion. Doing almost similar work.If you give a write up of complete procedure step by step & link will be of more help

[tebasuna51]

I'm unable to get that center_cut_GUI.Where can I get?

http://www.moitah.net/download/latest/Center_Cut_GUI.zip

[manolito]

I agree that the CenterCut method creates very good sounding 6-ch upmixes...

Just keep in mind that CenterCutCL.exe insists on 16int source files. If you feed it with a 24int or 32float source it will throw this error:

Center Cut CL v1.6.0
Copyright 2006-2010 J.D. Purcell
Center Cut algorithm by Avery Lee
FFT library by Takuya Ooura
http://www.moitah.net/

Error: WAVE must be PCM format.

[nivedhya]

http://www.moitah.net/download/latest/Center_Cut_GUI.zip

Got it Thank You.

[nivedhya]

Above Centercut GUI gives center & one side channel. This side channel is Front Left & Right Or Surround Right & Left. Got bit confused how to make surround from this? And LFE what to do? Some detailed explanation will be of help

[tebasuna51]

The side channels are Front channels, to create the Surround channels you need use other method, Center-Cut don't help you for that.

The LFE channel must be empty like I say you before (check the 'Bass to sides' in Center-Cut to let the low frequencies in front channels without changes).

To obtain the Surround channels there are many methods than involve some operations:

1) Mix the front channels with many coeficients values (at user taste):

SL = c1 x FL - c2 x FR
SR = c1 x FR - c2 x FL

with c1 + c2 = 1, c2 can be 0

2) Filter the frequencies (or not). For instance with a 100-7000 Hz bypass.

3) Delay the channels (or not). For instance with a 0.02 sec delay.

4) Phase shift the surround channels or not.

The tool to do these operations is your choice (Bidule, Audition, Nuendo, ...) I can help you only with free software (ffmpeg, sox, AviSynth, foobar2000, Audacity,...)

[manolito]

I think this little tool can get you started... :D

https://files.videohelp.com/u/172211/Upmix_CenterCut.zip

From the ReadMe:
Upmix_Centercut

This is a simple batch file to upmix a Stereo Wave file
to a (pseudo) 6-ch surround Wave file. It uses SoX, FFmpeg,
MediaInfo and CenterCutCL.

Extract all files from the archive into a folder, then open the
batch file with a text editor. You need to enter a few variables
right at the top of the batch file.

You can specify if you want to use an LFE channel (NOT recommended),
you need to enter a temp folder name, and a value for Peak Normalizing
is required. The bit depth of the 6-ch output file is always 32-bit float.

The batch file only accepts one parameter, and this is the source
Wave file which must be Stereo. You can also Drag & Drop the source
file on the batch file.

Please note that I use the -b parameter for CenterCutCL. This will
move the bass from the center track to the side tracks. In combination
with using an empty LFE channel this sounds very good to me.
But of course feel free to play with the batch file parameters...


Cheers
manolito

[richardpl]

The side channels are Front channels, to create the Surround channels you need use other method, Center-Cut don't help you for that.

The LFE channel must be empty like I say you before (check the 'Bass to sides' in Center-Cut to let the low frequencies in front channels without changes).

To obtain the Surround channels there are many methods than involve some operations:

1) Mix the front channels with many coeficients values (at user taste):

SL = c1 x FL - c2 x FR
SR = c1 x FR - c2 x FL

with c1 + c2 = 1, c2 can be 0

2) Filter the frequencies (or not). For instance with a 100-7000 Hz bypass.

3) Delay the channels (or not). For instance with a 0.02 sec delay.

4) Phase shift the surround channels or not.

The tool to do these operations is your choice (Bidule, Audition, Nuendo, ...) I can help you only with free software (ffmpeg, sox, AviSynth, foobar2000, Audacity,...)

ffmpeg have filter for upmix, it is slow, but working solution.

[tebasuna51]

@manolito
Your Upmix_Centercut.bat do the method with ffmpeg & sox.

But I can't understand this:
...rear_left.wav" remix -m 1v0.473,2v-0.473 delay 0.02
...rear_right.wav" remix -m 1v0.473,2v-0.473 delay 0.02
Now rear_left is the same than rear_right, maybe must be:
...rear_left.wav" remix -m 1v0.473,2v-0.473 delay 0.02
...rear_right.wav" remix -m 1v-0.473,2v0.473 delay 0.02

[EDIT]
BTW I don't like very much this kind of mixes for rear channels:
If there are a instrument I1 clearly in left channel, and other I2 clearly in right channel, we can listen I2 in rear_left and I1 in rear_right.
If it is like a echo maybe need a different delay.

Of course each user can select the method

[manolito]

Thanks Tebasuna for the comments...

I did not come up with these algorithms myself, I got them from this older post by Selur:
https://forum.doom9.org/showthread.php?p=1385573#post1385573
All the methods in this post have this rear channel method, so I thought it would be fine. I do not have a surround speaker setup, I still prefer plain old Stereo. So I cannot really test how these upmixes sound...

[tebasuna51]

Yes, in the Selur post there are many methods changing coeficients and other parameters.
The user can select any of them.

Like I say in other thread my preferred is https://forum.doom9.org/showthread.php?p=1640451#post1640451

But using only Center cut and ffmpeg you can do something like:
@echo off
rem Drag & drop any source
set SOURCE="%~1"
rem Paths to programs
set CENCUT=C:\Portable\0\CenterCutCL
set FFMPEG=C:\Portable\0\ffmpeg
rem
rem The source can be any input, source.wav is always supported by Center Cut
%FFMPEG% -i %SOURCE% -c:a pcm_s16le -ac 2 source.wav
rem Run Center Cut
%CENCUT% source.wav -s front.wav -c center.wav -b
rem Creating lfe empty
%FFMPEG% -i center.wav -af "volume=0" -c:a pcm_s16le lfe.wav
rem Create rear channels
%FFMPEG% -i front.wav -filter_complex "highpass=f=100, lowpass=f=7000, pan=stereo|FL = .47FL - .47FR|FR = .47FR - .47FL, adelay=delays=20:all=1" -c:a pcm_s16le -ac 2 rear.wav
rem Merge all channels and encode
%FFMPEG% -i front.wav -i center.wav -i lfe.wav -i rear.wav -filter_complex "amerge=inputs=4" -acodec ac3 -ac 6 -ab 640k -center_mixlev 0.707 output.ac3
pause

[manolito]

Thanks, I will give this ffmpeg based method a run right away...

BTW the link in your previous post does not work, it got truncated by copy&paste.

A few questions:
I noticed that you retrieve the rear channels from the CenterCut output while the methods I use creates the rear channels from the stereo source file. This will make a slight difference. Is it important, and which one sounds better?

Since CenterCutCL only works in 16-bit your method keeps everything at 16-bit the whole way through. You also do not normalize anything during the whole procedure.

I remember that when I implemented the SoX based methods in AVStoDVD, I had to take great care to avoid any clipping. I had to use normalizing at several stages, and still the only way to avoid any clipping in the output was to use 32-bit float for the intermediate WAV files. Since your method does not need intermediate WAV files, is normalizing unnecessary with ffmpeg? And does ffmpeg process the audio internally with float precision automatically?


Cheers
manolito

[SeeMoreDigital]

I wonder what happened to Moitah (https://forum.doom9.org/member.php?u=51417)?

[tebasuna51]

BTW the link in your previous post does not work, it got truncated by copy&paste.
Yep, sorry, updated the link.

I noticed that you retrieve the rear channels from the CenterCut output while the methods I use creates the rear channels from the stereo source file. This will make a slight difference. Is it important, and which one sounds better?
Maybe it is not important for music, but for a movie the main dialogs are extracted to center channel and deleted in fronts.

If we use the source to create the rear we listen the dialogs delayed by the rear channels, and all the movie sound with dialog echoes, like filmed in a cave.

Since CenterCutCL only works in 16-bit your method keeps everything at 16-bit the whole way through.

It is a CenterCut limit, after any downsize to 16-bit we can't recover a better precission, upsize the samples is usseless.

You also do not normalize anything during the whole procedure.

CenterCut extract by default at 100% volume the front and center channels unchanged in the procedure.

To control the rear volume you have the coeficients mix (must be always c1 + c2 <= 1).

BTW if the source need to be amplified you can do it at the beginning.

To do the frequency filters I think ffmpeg convert the samples to float but we can't expect any important clip in the process.
I think than normalize is unnecesary.

[manolito]

:thanks:

[manolito]

After some testing I have a few findings and questions... :)

First of all the ffmpeg filter options for creating the rear channels must be quite new. My standard ffmpeg version for WinXP is from 2016, and this version does not work. I had to upgrade to the latest Reino version from 2020, and this fixes it (but also brings some other problems). But what the heck...

Next thing I found out is that staying in 16-bit throughout the procedure is NOT safe to prevent clipping. I used an input clip in the Opus format which had tons of clipping after decoding it to 32-bit float WAV. The audio is from the clip in this post:
https://forum.doom9.org/showthread.php?p=1926825#post1926825

Before feeding it to the upmix routine I normalized it to -2.5 dB, but this was not enough to prevent clippig in the rear channel creation command. Converting the CenterCut output files to 32-bit float took care of this problem.

Another issue is that the center-mix parameter in the final mixing call seems to be a private option for the AC3 encoder which is invalid when converting to a multichannel WAV file.

https://i.postimg.cc/Rq5wSxh9/Center-Mix.png (https://postimages.org/)

Why are you reducing the center mix level at all? Do you feel that Moitah got his defaults wrong? CenterCutCL does not offer an option to attenuate the levels for the center channel and the side channels separately, so the only way to just attenuate the center channel would be to call it twice. Or is there an ffmpeg parameter to reduce the center level for WAV output?

Whatever, I think your routine is worth to be included in my AVStoDVD plugin. Having alternatives is always a good thing.

[tebasuna51]

First of all the ffmpeg filter options for creating the rear channels must be quite new. My standard ffmpeg version for WinXP is from 2016, and this version does not work. I had to upgrade to the latest Reino version from 2020

Is good have the soft updated. Congratulations.

Next thing I found out is that staying in 16-bit throughout the procedure is NOT safe to prevent clipping. I used an input clip in the Opus format which had tons of clipping after decoding it to 32-bit float WAV.

If the source have already the clip (like your source have) you can do nothing to recover the original sound. Normalize that source is usseless, the clips remain.

Another issue is that the center-mix parameter in the final mixing call seems to be a private option for the AC3 encoder which is invalid when converting to a multichannel WAV file.

Of course, is a option of the AC3 encoder ignored when you output wavs. My code is only a sample of use.

Why are you reducing the center mix level at all?...

Just the opposite, the default Center Mix Level in ffmpeg ac3 encoder is 0.595, and I force to 0.707. (only 3 values are allowed these and 0.501)

BTW this metadata stored in AC3 headers is used only when a AC3 decoder is forced to downmix a 5.1 to 2.0. When the decoder output 5.1 this parameter is not used.

[richardpl]

Why you guys use subpar upmixing tehniques?

[manolito]

Why don't you just post your "superior" upmixing technique instead of posting destructive comments?

[tebasuna51]

Why you guys use subpar upmixing tehniques?

This thread is 'a stereo (music) to 5.1 with Moitah's CenterCut...' then we try to explain to the user how use it with free soft (ffmpeg).

I also put my preferred method to 2.0 -> 5.1.

Please do the same and put your method.
If your suggestion is use ffmpeg with the surround filter (https://ffmpeg.org/ffmpeg-filters.html#surround) please open a new thread and explain how to use the 48 parameters and the 20 functions for win_func.

[EDIT] I can add your method to 'Sticky: GUIDE LIST: Stereo-to-Surround Conversion Guides (https://forum.doom9.org/showthread.php?t=83752)'

[manolito]

Thanks for bringing up this new FFmpeg Surround filter, I was not aware that it existed...

I made a few tests (only disabled LFE creation, otherwise all default settings), and it sounds pretty good. No idea if the center channel can compete with the CenterCut method.

Since I cannot listen to the upmix results on a surround system I uploaded 3 test results using either CenterCut or the new FFmpeg Surround filter.
https://www.sendspace.com/file/n8yxg0

The source is again from this post:
https://forum.doom9.org/showthread.php?p=1926825#post1926825

Since I can only judge how these upmixes sound on my stereo speakers (downmixed by the Windows Sound Mapper) I invite everyone to play these upmixes on a good sounding surround system and let me know what you think.


Cheers
manolito

[nivedhya]

I think this little tool can get you started... :D

https://files.videohelp.com/u/172211/Upmix_CenterCut.zip

From the ReadMe:



Cheers
manolito

You have mentioned as Pseudo multichannel which I hope something like prologic ie: same stereo front left & right being played over surround speakers too.If it is that my AVR directly plays stereo in multichannel stereo mode & I don't think we have to struggle this extent for it. Kindly let me know is it that multichannel or something else

[manolito]

same stereo front left & right being played over surround speakers too.

No, it is not like this... :eek:

Of course if you only have two front channels as your source the signals for the surround speakers have to be generated from the source channels in some way. But this involves some processing of the source channels. Tebasuna has outlined the basic principles here:
https://forum.doom9.org/showthread.php?p=1930809#post1930809

There are different methods to process the signals, and this is the reason for the different sounding upmixes. Please download my conversion results from my previous post, play the files over a surround system and pick your favorite method...

And of course you can question the usefulness of all this upmixing stuff because your AVR can do the same with the help of a DSP (Digital Signal Processor) which is in hardware and can do the upmixing in real time. The drawback of this method is that you are at the mercy of the chip designers who designed the upmixing algorithms.

[tebasuna51]

Since I can only judge how these upmixes sound on my stereo speakers (downmixed by the Windows Sound Mapper) I invite everyone to play these upmixes on a good sounding surround system and let me know what you think.

First of all please put the ffmpeg command line.
I make some previous test with:
-af "surround=lfe_out=0:level_out=2"

Say you also if you use the same source level (-2.5 dB) for all test.
Like you can see in the image there are volume differences.

We can use a sample with more quality than this over saturated opus audio to do more test.

In the image you can see also:

- The front channels in CenterCut upmix seems have more dynamic range than ffmpeg upmix.

- The center channel in ffmeg upmix have a excesive volume compared with the rest.

- The surround channels have low volume in ffmpeg upmix.

Of course this can change with the source, and parameters used in ffmpeg surround filter. And with user preferences.

[EDIT] I don't see big differences between the CenterCut ffmpeg and Sox upmixes.

In this sample the volume balance between channels in CenterCut upmix is my preferred

[tebasuna51]

@nivedhya
Seems you mix the conceps like manolito try to explain you.

Create 5.1 from 2.0 is not recommended at all.
Is only usefull when you want create a unique multimedia file with sources 2.0 and 5.1 and want add them.
Or only to experiment.

To listen a stereo audio with a 5.1 AVR you have the options:

1) If you know the 2.0 have DPL compression activate the DPL function in the AVR. Some AC3 files must activate the DPL by default. Then you have a 5.1 similar to original.

2) Listen the audio 2.0 like was created (recommended)

3) Use your AVR DSP functions like say:

And of course you can question the usefulness of all this upmixing stuff because your AVR can do the same with the help of a DSP (Digital Signal Processor) which is in hardware and can do the upmixing in real time. The drawback of this method is that you are at the mercy of the chip designers who designed the upmixing algorithms.

4) Create you own 5.1 with these methods, for me it is waste time and space, unless you try differents methods for each source until obtain the perfect upmix for you.

We only can show you the methods but never we can say what is the better for all sources.

[manolito]

First of all please put the ffmpeg command line.
I make some previous test with:
-af "surround=lfe_out=0:level_out=2"

I simply used
-af "surround=lfe=0"
because this option appears right at the top of the documentation. Otherwise everything was at the default.

My workflow was as usual:
1. Get the source from AviSynth as a 32-bit float WAV.
2. Normalize to -2.5 dB.
3. Apply the upmix algorithm and output a 6-ch 32-bit WAV.
(For CenterCut convert the source to 16-bit first, apply CenterCut and reconvert the CenterCut output to 32-bit again)
4. Normalize the 6-ch output to -1 dB.
(As an alternative use EBU R128 normalizing using either BS1770gain or LoudNorm, target LUFS -18 dB). Not used in my tests.
5. Convert to 448 kbps AC3 (for DVD use)

Listening to the results on stereo speakers I also noticed that the FFmpeg Surround result was a bit louder than the CenterCut results, even though I had used the same final normalizing to -1 db. Must be caused by the louder center channel.

Thanks for your tests,
Cheers
manolito

[Richard1485]

As an experiment, I'm adapting Tebasuna's method to create a mono surround instead (L,R,C,B).

Stereo surrounds:

ffmpeg -i front.wav -filter_complex "highpass=f=100, lowpass=f=7000, pan=stereo|FL = .47FL - .47FR|FR = .47FR - .47FL, adelay=delays=20:all=1" -c:a pcm_s16le -ac 2 rear.wav

Mono surround:

ffmpeg -i front.wav -filter_complex "highpass=f=100, lowpass=f=7000, pan=mono|c0=0.5*c0+0.5*c1, adelay=20" -c:a pcm_s16le rear.wav


I don't think the result is too loud, because the coefficients are equal to 1. Ideas for further experimentation would be welcome.

[tebasuna51]

Everybody can experiment of course.

But, like in video I don't recommend upsize the resolution, also in audio I don't recommend upsize the audio channels.

In TV's there are upsize routines to fit the screen from the original input, in audio amplifiers/receivers there are DSP functions to use the all available speakers with many effects.
My Denon have "Rock Arena", "Jazz Club", "Matrix", ...

With the BackCenter = (.5FL + .5FR).Delay(20ms) maybe we obtain only a echo, I don't know if is enough to listen it like surround.

[j7n]

Just keep in mind that CenterCutCL.exe insists on 16int source files. If you feed it with a 24int or 32float source it will throw this error:
24-bit and 32-bit integer work fine and maintain precision. The wav files must have codec ID 1, as written by old software, not "extensible" ID -2. Set it with a hex editor or WaveWizard.

[Richard1485]

But, like in video I don't recommend upsize the resolution, also in audio I don't recommend upsize the audio channels.

I don't recommend it either. It's just that in this case I have my reasons for experimenting. :D
With the BackCenter = (.5FL + .5FR).Delay(20ms) maybe we obtain only a echo, I don't know if is enough to listen it like surround.
That's the real problem. You are right. It is like an echo, but maybe good enough for some uses...
24-bit and 32-bit integer work fine and maintain precision.
That's right. There's a note in the changelog about support for "8/16/24/32 bit audio". 24-bit wav works fine for me after eac3o input output -simple to make an input file.

[manolito]

@ j7n

The wav files must have codec ID 1, as written by old software, not "extensible" ID -2.

This is not true. 32-bit integer is not a format which is used for WAV audio, for 32-bit only float is used. I wrote a 32-bit float wav using either the ancient WaveLab3 software, or I used Wavi to extract the WAV from the old AVS 2.60. Both tools do not support the "extensible" wave format. And in both cases the resulting float wav has the codec-ID 3. And CenterCutCL refuses to accept these float WAV files as the input.

I did not try if changing the codec-ID to 1 using a hex editor would help.


@ Richard1485

You are probably talking about the ffmpeg changelog. CenterCutCL does not come with any changelog which says that 24 and 32 bit audio is supported. Please read the posts you are replying to...

[Richard1485]

You are probably talking about the ffmpeg changelog. CenterCutCL does not come with any changelog which says that 24 and 32 bit audio is supported. Please read the posts you are replying to...

I do read the posts to which I reply. If I make a mistake, it's not because I'm not reading. Of course I'm not talking about ffmpeg's changelog. I'm talking about moitah's changelog for the CenterCut plugin. I've tried CenterCutCL with 24-bit audio. It works, and the output is 24 bit. I haven't tried 32-bit audio yet (and didn't say that I had).

[manolito]

I'm talking about moitah's changelog for the CenterCut plugin

Would you care to tell me where in the CenterCut download from this post
https://forum.doom9.org/showthread.php?p=1929561#post1929561
you found a changelog?

[Richard1485]

Would you care to tell me where in the CenterCut download from this post
https://forum.doom9.org/showthread.php?p=1929561#post1929561
you found a changelog?

Would you care to tell me where I said that I found a changelog in the download?

EDIT: And I downloaded after following the link in the OP.

[manolito]

You are probably talking about the changelog on moitah's site about the Winamp DSP version of Centercut. This is not the Centercut version we are talking about here.

[Richard1485]

You are probably talking about the changelog on moitah's site about the Winamp DSP version of Centercut.

Yes. That's the one that I'm talking about, the Winamp plugin.
This is not the Centercut version we are talking about here.

I know, but presumably moitah based his later version on the Winamp plugin. The CenterCutCL version that I'm using certainly does accept 24-bit audio. I haven't tried 32-bit audio yet, but I will later.

EDIT: 32-bit int works for me on CenterCutCL.

[j7n]

This is not true. 32-bit integer is not a format which is used for WAV audio, for 32-bit only float is used.I wrote integer. The format exists, and is the native internal format of SoX, and can also be saved from Sound Forge. There is usually not a practical need to use it, but if float is not available, one could reduce the volume considerably to avoid clipping.