Split channels of a dual-channel MP3 without re-encoding

[zuja9]

Hey there,

I have a bunch of MP3s that were encoded (recorded) in dual-channel mode (so it's not mono, stereo or joint-stereo). According to the specifications of the MP3 standard, this means that there are two completely different audio tracks (with half bitrate) within the MP3. If they really are like separate mono tracks put next to each other, I would like to export them into two separate MP3 files without re-encoding anything. I hope you understand what I want

I thought mp3DirectCut was able to do it, but I was wrong.
Now I'm looking for a proper tool (even command-line tools are perfect for me).

Thank you,
zuja9

[Cornucopia]

AFAIK, there is NO software that does this (nor the reverse). Wouldn't be impossible for it to be made, though.

Sometimes "stereo" tracks are still encoded as dual-channel (unusual, but happens, particularly when the 2 channels are VERY different). In that case, you WOULDN'T want to change things.

Otherwise, you WOULD have to re-encode.

Scott

[zuja9]

There should be some tool at least a command-line one. My problem is that these recordings are mono but stored as stereo. I took an MP3 like this and compared its left channel to its right by inverting one of them (using Audacity). Almost no difference, only some static noise. So it would be great to strip down the size of these recordings by just keeping one channel of the two. I have gigabytes of these MP3s.

[jman98]

You are being vague, in my opinion, about what you have. One possibility is that this is some kind of VCD conversion from a typical Hong Kong VCD with Cantonese on one side and Mandarin on the other. I guess it could be music as true 2 channel (duophonic) audio is certainly possible under MP3 and other audio formats and it's not difficult to find examples of such (ie. RCA orchestral recordings from 1954 and 1955 such as the Chicago Symphony Orchestra under Fritz Reiner, the first 2 UK Beatles albums) but as to why you would want to have that kind of musical thing separated is something I can't answer.

Anyway, I think Audacity can let you save just one channel of the MP3 into mono. I cannot answer whether it will have to re-record or not, but honestly, you're not exactly looking at a million different options here so 1) you are probably just going to have take what you can get 2) can you REALLY tell a difference it if does re-encode?

[zuja9]

I'm not vague, I know what I want I would like to demux/extract the channels of an MP3 in which they are stored separately, and save them as two separate mono MP3s. According to MP3 Tech's specification, in stereo and dual-channel mode channels are stored as two different audio streams. The only thing they share is the half of the bitrate.

Well, I know Audacity can separate stereo channels, but it can't do it without re-encoding. This is important for me. It would take a lot of time to re-encode all my recordings, let alone the quality loss that would be caused by it.

I have found a tool that can do the same with .mp2 files, but I can't get it work with MP3s
It's called BeSplit and has different versions all across the net.

Latest stable version v0.82 won't do anything.

d:\>besplit -core( -input REC001.MP3 -prefix mt -type mp3 -demux )

BeSplit v0.82 by DSPguru.
--------------------------

[00:00:00:000] +------- BeSplit -----
[00:00:00:000] |  Input : REC001.MP3
[00:00:00:000] |  Output Prefix : mt
[00:00:00:000] +---------------------

Latest available unstable version exports two zero-byte .mp2 files
(it seems to ignore the command line option -type mp3).

d:\>besplit -core( -input REC001.MP3 -perfix mt -type mp3 -demux )
BeSplit v0.9b8 by DSPguru.
--------------------------

[00:00:00:000] +------- BeSplit -----
[00:00:00:000] |  Input : REC001.MP3
[00:00:00:000] |  Source Sample-Rate: 32.0KHz
[00:00:00:000] |  Channels Count: 2, Bitrate: 128kbps
[00:00:00:000] |  Output : mt
[00:00:00:000] +---------------------
[00:00:00:036] +---------------------
[00:00:00:036] Operation Completed !
[00:00:00:000] <-- Process Duration


Visit BeSplit's Webpage at :
                                http://BeSplit.doom9.org/

So I think I'm not that far as I used to be. Any ideas how to get this tool work?

[jman98]

I guess it's not our business, but yes you ARE being vague. Why would someone create a file, which is as you put it, "two different audio streams"? That's nonsense, Well, I provided one reasonable explanation for it. But again, you don't wish to talk about exactly why someone would create such a file in the first place, so that just makes us all wonder exactly what you have and why you want to do this. It's kind of like saying "I need the name of a surgeon because I want to have my left foot amputated" without providing the context that your foot has cancer. But whatever.

As far besplit goes, what OS do you have? Is it 64 bit? Also, you have 32 KHz audio? Dude, that is gonna suck! Anytime some idiot uses a sample rate under 44.1 KHz, you are going to have audible artifacts in the sound.

[nic2k4]

You probably see a difference between the 2 versions only because you mistyped -prefix. If you just type besplit on the command line, you will see that the demux option is only for wave files. I don't think it's possible to demux an MP3 or "split" its channels without re-encoding as the audio data is recorded in frames, unlike a WAV file with its audio data recorded in PCM. If you really must do it, try BeSweet in GUI mode and set it to convert 1 channel to mono. There is a batch mode too. I would try it on 1 file and see how much time it would take and if the space savings are worth it.

[Cornucopia]

Things to try:

1. Contact BeSplit's author: DSPGuru. He might lurk here, but is probably over at Doom9. If that isn't part of the app, you still might be able to use the source code...
2. Try reencoding anyway. If a source was, say: 2ch - 16bit - 32kHz - 128kbps, you could probably get by splitting it into mono and encoding: 1ch - 16bit - 44.1kHz (or 48 or even 32) - 64-96kbps.
3. Leave things alone and live with the size.

Scott

[zuja9]

I don't understand why I have to give an exact reason for everything. I just asked for a software which can do a technically possible thing, but if you want to know I have a voice recorder that makes recordings in the format I described. I have a lots of recordings like this. I don't need alternative workarounds just want to do what I wrote in my first and second post: split these channels without re-encoding and output them into mono MP3 files. I'm not willing to be rude, but I need to tell: if you know how to do it, please help me, if you don't know, please don't waste your time calling me "vague" or question the necessity of the things I want to do.

Contacting the author of BeSplit was a good idea, thanks. I will.

[tostada]

I guess it's not our business, but yes you ARE being vague. Why would someone create a file, which is as you put it, "two different audio streams"? That's nonsense, Well, I provided one reasonable explanation for it. But again, you don't wish to talk about exactly why someone would create such a file in the first place, so that just makes us all wonder exactly what you have and why you want to do this. It's kind of like saying "I need the name of a surgeon because I want to have my left foot amputated" without providing the context that your foot has cancer. But whatever.

He was slightly vague in the first post. He neglected to mention that both channels are identical until his second post. He said his files weren't stereo, which is confusing, because they are stereo recordings of a mono source. But you seem to have completely ignored that second post which should've cleared everything up.



Why would someone create a 2.0 mono file where both channels are identical? Maybe because the file was made with software that processes everything in stereo even if the source is mono. Maybe because some recievers output a 1.0 source through just the center channel and it sounds better coming through the left and right speakers. I don't know.

But the fact is that there are a LOT of 2.0 double-mono files out there, and if they're not encoded in joint stereo then half the bitrate is completely wasted.

This is quite prevalent with DVD sources. If a DVD is mono then it is usually recorded in 192k 2.0 AC3. I've ripped hundreds of DVDs, and I don't think I've ever seen one that actually had a 1.0 AC3 track. I don't know if true 1.0 is even compatible with the players or the DVD specification. So it should be simple to split the 192k AC3 track into two 96k tracks and just get rid of one of them, so you have half the bitrate but absolutely no loss in quality.

edit: Funny. Right after I posted this I ripped an old copy of Barfly (1987), and it has 1.0 AC3 audio.



When I rip a DVD I always include the original 2.0 or 5.1 audio. It doesn't take up much more space than transcoding it to a slightly lower bitrate of MP3 or AAC. But if it's mono, it's kinda silly just artificially bloating your filesize.

As much effort as various communities spend on squeezing every last bit of quality out of audio and/or video encodes, I'm shocked that there is nothing out there to be found that can split the channels in stereo MP3s or 2.0 AC3 or other non-joint stereo formats. BeSplit is the only thing I've been able to find, and it can only demux WAV and MP2.

There are several threads on Doom9 about this, and they tend to get a lot of "why would you want to do that?" replies, or confused replies by people who don't understand what the question is. I'm surprised this hasn't been solved in 10 years. Well, it has been solved for MP2s, which is mostly only applicable to the already mentioned case of HK VCDs with different languages on the left/right channels.

[szciklon]

Hi,
I have a similar problem. I have stereo MP3 files, recorded lectures/workshops, one channel is English, the other is Hungarian (sometimes, when the speakers change the translation also shifts). I would like to have two mono mp3 files. I hope it is clear what I would like to achieve...

[hello_hello]

I guess it's not our business, but yes you ARE being vague. Why would someone create a file, which is as you put it, "two different audio streams"?

I recall many times, reading advice in forums not to use joint stereo when encoding MP3s, because the person who offered the advice was under the false impression it lowers the quality. Maybe someone encoded a bunch of files which were mono or "dual mono" or they were converted from mono to "dual mono" when encoding, and for some reason joint stereo wasn't used.

zuja9,
Are you 100% sure joint stereo wasn't used when encoding these MP3s? If it was, they'd probably use no more bit rate than just one mono channel on it's own.

The trouble is 99.999999999999% of MP3s, whether they be "dual mono" or stereo, will have been encoded with "joint stereo" enabled, which means the two channels aren't independent, so that's probably why no software (I know of ) exists to separate the two channels as most of the time even "dual channel" mono MP3s would be inseparable.