why does it do this:

PCutMP3 -- Properly Cut MP3 v0.95b by Sebastian Gesemann

scanning "Essential Mix - Mylo - 10-APR-2005-2.mp3" ...
first frame header = MPEG1 Layer3 192kbps 44100Hz Stereo
Xing/Info tag present
bitrate = 192 kbps (CBR)
accurate length = no
157849344 samples (is NOT a multiple of 588)
writing "Essential Mix - Mylo - 10-APR-2005-2-01.mp3" ...
Exception in thread "main" java.lang.ArrayIndexOutOfBoundsException: -10
at de.zebee.mpa.ScannedMP3.getInt16(ScannedMP3.java:85)
at de.zebee.mpa.ScannedMP3.getInt32(ScannedMP3.java:89)
at de.zebee.mpa.ScannedMP3.getFrameFileOfs(ScannedMP3.java:103)
at de.zebee.mpa.ScannedMP3.crop(ScannedMP3.java:329)
at de.zebee.mpa.MainCLI.main(MainCLI.java:270)

Hmmm... looks like pcutmp3 is choking on your mp3 because it has already been cut by another tool. another tool which made the first franes undecodable. I'm not sure but I thought I tested pcutmp3 on those streams, too. Could you upload the first couple of kilobytes of this mp3 file and the CUE sheet anywhere ? It'd make things easier for me.

Sebi

first 10 seconds of the mp3: http://rapidshare.de/files/13607490/Essent...2005-2.mp3.html
cue: http://rapidshare.de/files/13607522/Essent...2005-2.cue.html

It makes sense that this was cut, because it's the second half of a 2 hour long radio mix that someone split so it would fit on 2 cds. I already tried rebuilding the header using foobar2000, but that didn't work.

A quick fix that might work: Fix the bit reservoir errors using mp3brfix. Actually pcutmp3 is supposed to fix these errors as well. But it didn't seem to work in your case for some reason. I'm going to check that.

later added:
If you have the first half you can try to reconstruct the original stream by
copy /b half1.mp3+half2.mp3 full.mp3
without applying mp3brfix on the 2nd half first.
This way you won't loose the start of the 2nd half due to the bitreservoir errors.


Sebi

Unfortunately i wasn't able to download the files you linked because someone already did it and the hoster only allows one download. Please PM me a link.

Sebi

is there any disadvantage of cutting a mp3 twice using this tool? Let's say I split a large file and then do the fine tunning? With disadvantage I mean for instance a higher enc_delay or enc_padding since that value is not interpreted correctly by all music players.

thanks

No. There shouldn't be a problem.

Sebi

I use XBMC quite frequently and I noticed that it did not play tracks that had been split with pcutmp3 properly; it missed a few seconds off the start of each track. So I posted it as a bug, which they fixed, but the problem was really with the mp3 files themselves. Here is my bug report, note in particular the reply

Is this something that needs fixing? Everything I play the files with works for me now, but I was just bringing up the problem.

I don't think Sebi is actively developing this great tool at the mo / anymore. He made the source available for others to use / update / tweak, but I don't think anyone has. Shame, cos it's a great tool.

dignick: Did you try the latest version? I thought I already fixed the seektable generation (see post#4)

The only pcutmp3 bug I know of is it may choke on previously "unproperly" cut files.

BTW: I was thinking of doing a partial rewrite when I've more time on my hands. The code where all the magic takes place (ScannedMP3.java) looks kinda messy currently.




Can you elaborate? I can't believe that incorrect seektables would cause the behaviour you described. It's probaby the usually higher "ENC_DELAY" values from the LAME header that cause XBMC to do weird stuff.

Sebi

yep, iTunes does the same. I've cut a song where there is silence up in front, but iTunes plays the last few frames before the silent part. foobar is getting ir right though

I have been using 0.94b, now I see the new jar is 0.95b, but you havn't updated the link title which still reads 'direct link to version 0.94 beta' so I didn't know there was a new version. This has possibly fixed the problem but I can't check it as everything works anyway. Shame I've been cutting all my mp3's with an old version though.
I could give you an mp3 that has been cut with the two different versions if you want. I don't know in any more depth what the problem was with the files not playing properly, just that they weren't before it was fixed, and they are now. If you could be bothered you might be able to find out from the xbmc cvs, it could have a reference to my bug report with respect to the changes made.

A small feature request would be to replace forbidden characters with a default character like _ or -, as I get errors a lot with this and I have to fiddle with the cue every time.

Please keep developing this sebi, your work is much appreciated!

This is simply iTunes not handling the LAME tag. As noted erlier the use of this tool is limited. If you mainly use players that don't support the LAME tag it doesn't make much sense to use pcutmp3.

I thought that by dignick meant the player skipped a few seconds of the start. The iTunes behaviour seems to be unrelated then.

BTW: I think the 0.95b version fixed the "illegal character" thing.

Sebi

SebastianG> Would it be too difficult for you to make a GUI version? Maybe something in LameGrop/OggDrop style. Just a very simple GUI, which would make life easier for a lot of people.

Maybe I can try to do it myself, I already started learning Java 2 times, but I never got enough time to produce anythig really usable. And then I forgot most of what I've learned.

Just tried to cut a 48000hz mp3. It cut at the wrong times, just as it thought it was cutting a 44100hz mp3.
Is there a workaround for this?
Thanks for making this great tool anyway!

Will pCutMP3 ever have a GUI? Command line is powerful for batch work, but what about ocassional splits where we don't know the frame number?

how did you end up with those mp3's in the first place


I have a cue sheet that has three tracks in it. That way I can cut something from the end as well as from the beginning. I use foobar to listen to the cue sheet and then adjust the times on the cue sheet till I get the result I want. Works quite well

Just out of curiosity: How did you specify the time?
(cue sheet, in samples, mm:ss.xxx?)

I believe the time in (CDDA-)frames from the cue sheet is directly multiplied with 588 samples (regardless of the sampling rate) which may explain the behaviour you've encountered. The time given in mm:ss.xxx should work as expected. (I may be wrong -- havn't verified it lately)

Sebi

1) Any pro audio source - hard disk recording, DAT, etc.
2) DVD rip
3) Recording at 48kHz since he has a Soundblaster or other card that forces onboard resampling to 48kHz

There are MANY other sources in the world for digital audio than just CDs...

I used this cuesheet on a 48000hz mp3 file:

...which gave me this result. To me it seems that pcutmp3 thinks this is a 44100hz file which could explain the strange behaviour. I wonder if this is similar with other samplerates.

My favourite band Underworld did some "download-only" releases lately. Their last one, which was released on Monday, was accidently encoded at 48000hz. They have fixed that in the meantime though.

It correctly detects the sampling rate but the CDDA-frame numbers given in "MSF"-format in a CUE sheet are just translated to sample offsets by multiplication by 588. All other methods of specifying the time (non-CUE sheet mode) should work as expected. A quick fix should be easy (replacing 588 by samplingrate/75 in the source code).

Sebi

Hi Sebi. First of all, thank you for this tool!

Have you considered working any more on the "rejoin" feature you hinted at in your initial post?
Here is the thought process which prompts the question:

If someone has a live recording (picture an outdoor festival...) that was encoded with L.A.M.E., and which plays gaplessly in players that support the L.A.M.E. tag, it should be possible - assuming one can losslessly rejoin those MP3s - to enclose the rejoined MP3 in an MP4/M4A container, add chapter stops, and play the result gaplessly on an iPod.

Yes, I realize Rockbox can play this source gaplessly without having to enclose anything in an M4A container... but this is for my wife's iPod. She is a Mac fanatic, she uses iTunes, and I'm not about to ask her to switch.

- M.

"rejoining" will only work if the mp3 stream has been split using pcutmp3. It won't do the trick if you have seperately encoded MP3 files (even when they play back gaplessly). Regarding wrapping it into an MP4 container: I simply don't have the resources to do that nor I'm willing to invest too much time on a feature that will hardly be used by anyone. Sorry. TBH, I very rarely use pcutmp3 itself which kind of explains the lack of motivation to improve it any further for now. It's in a usable state and serves its main purpose. However I do have a little TO-DO list for pcutmp3 which I intend to work through (rewrite of the core, enabling rejoining (to mp3))

Sebastian

Oh... well, thank you for clarifying. Guess I somehow assumed you'd found a way to concatenate a L.A.M.E.-encoded set and eliminate the encoder gap.

- M.

Sebi: if you think you can rejoin only pcutmp3 files, then could the following be done:
- use pcutmp3 to (quoting you here) "mark the silence frame (which only carries data in its main data section) as such and include some meta information. This allows rejoining of files automatically (in the future)"

What I'm getting at is: apply pcutmp3 to 'normal' mp3s (using some sort of samllest 'cut' time frame or maybe just the bit that marks the silence frame [and anything else that's required]) and then they should be available to rejoin.

I was just trying to apply a bit of lateral thinking to it. Of course, getting pcutmp3 to do this might not be possible.

How do I correctly compile pcutmp3 to an executable exe working on all machines even without java? I tried compiling with Excelsior, and it works on any PC with java installed, but yesterday i got an error telling me to use Jet II or something like that.

Please consult the customer support of xlsoft or have a look at this page -- it may be of help.
Since I don't use the "Reflection API" or load *.class files manually/dynamically native compilation should be possible in theory (otherwise it'd be somewhere between tricky and impossible). But I never tried any native Java compiler before and can't help you on this one any further.

edit: this is part of the Excelsior JET FAQ:

So, it seems that the compiler outputs a small EXE file that needs some DLL libraries (the Excelsior JET Runtime). So, if you copy just the EXE file to another computer which doesn't have the Excelsior JET Runtime libraries you can't run the application. -- just my interpretation


Sebastian

Damn, I always forget to read the small words... Sorry then

The problem is that MP3s still are made out of an integral number of frames. Sample accurate cutting through pcutmp3 only works by keeping an integral number of frames and signalling the decoder how many samples at the start / end to discard. Joining two (accurate-length-) streams (without introducing a little delay or a skip at the boundary) only works on a special condition regarding enc_delay/padding of the preceding track and enc_delay of the following track. Since 1152 (samples/mp3-frame) is not a multiple 588 (samples/cd-frame) chances are very low for seperately encoded MP3s that they can be joined "properly".

S

Thanks for the reply Sebi. It was just a thought that's all.

I found out that pcutmp3 throws an exception if it tries to write files that contain characters that cannot be used in filenames (< and > on Windows for example).

It doesn't with the newest version. Although sebi hasn't changed his page 1 post, the version it links to is actually 0.95b, which fixes the problem and simply removes the characters.

Thankyou very much for this tool, Sebastian. Does exactly what it says on the box, and does it perfectly.

In case anyone else is looking for cuesheets to match their albums, Cuesheet Heaven worked for me.

The program has worked great, but the non-ID3v2 support has been a pain at times. Sebastian have you found a library that you would like to use?

Sorry if it sounds like I am bugging you, it's the only drawback that is keeping this program from being PERFECT!

Hello guys,

I am just wondering if anyone has used this tool and then listened to the new (single track) mp3's with itunes 7? I am just curious to see what anyones results are.
I tried this morning and it seems that there is a very minor click (almost seemless) with the new mp3's, alot better than any other tool I've used.

If anyone has got any better results I would love to hear how you achieved them.

Also if anyone has any info on manually changing some settings (with pcutmp3) I would love to be a guinea pig and try them out.

Just for info the latest Ipod also had this very minor clicks.

Cheers

Rochey

Sorry, but todays is not my day...

Could anyone be so kind to post a batch file, for putting it into the "Send to" folder to execute automatic the pcutmp3.jar file?
All my tries with giving the parameters to the batch file via send to... ended up in a mess.

THX

Hmm, Codepage Problem on the morning...

Here comes a batch file for the "Send to" Menu.
Select File -> Send to -> Batch File
Customize for your one needs.
Output will be the source folder of the cue file.

This might be a dead topic, but I thought I'd give it a try anyway...

I've been quite pleased with pcutmp3. I've since re-split all of my files originally cut with mp3splt and smile every time tracks transition seamlessly being played back in foobar.

I own a Squeezebox and haven't been so lucky. Their newest version 6.5.1 with firmware 57 purports to support gapless MP3 playback using LAME headers, but I haven't been able to get it to work. Has anyone else experienced this problem?

Any help or advice would be greatly appreciated!

From SebastianG's original post:
emphasis mine

My guess is that the Squeezebox (as well as iTunes, in reference to an earlier post) can't deal with the unusually large delay values, just as Foobar2000 0.8.3 couldn't.

I'm no expert on the details of how this program works, but I do know that it isn't used by more people because it's a command line app. I wrote a GUI for it so you can tell your friends that they don't have to fiddle with XP and Vista's annoying command line, just double-click and it should work. All the backend stuff is the same, and if Sebastian releases new versions in the future it should be a simple file swap.



Right now it only supports CUE splitting. There is no overriding artist names or custom split points. Those will be implemented later. It also requires Java 6 for the time being. Sorry, but the class I used to implement multi-threading is Java 6 exclusive and it didn't say that anywhere. Thanks, Java docs! If you don't want to upgrade, I'll forgive you.

So you load the cue file you want. The program is smart and will look for the mp3 file. If it finds it, it fills it in so you don't have to browse for it. It defaults the output directory to the same directory the cue is in so you may want to be wary of that.

This is like the second revision I made, so I'm sure it's a little rough yet.
Download here

Feedack is always appreciated.

Nice and simply GUI. I actually wrote one just like this, as a Java/Swing practise, should have released it really. Recently I've just been using a batch file listed above

Very useful info about patching Foobar 0.8.3. This "value too big" error has been bugging me for a while, almost to the point that I installed 0.9 just to do gapless correction (but it didn't work anyway). Thank you, Sebastian.

While we're here, it's handy to patch 2 more offsets in order to enable manual entry of large delay and padding values.

80 04 00 00 8B D8 89 55 -> FF 0F 00 00 8B D8 89 55     (entry)
00 09 00 00 7F 04 3B CB -> FF 0F 00 00 7F 04 3B CB
80 04 00 00 33 C0 2B 5D -> FF 0F 00 00 33 C0 2B 5D     (what to do if greater than allowed)

thanks. is there any way to get a .exe ?

Just create a batch file, which is a text file with a .bat or .cmd extension and insert the following line:

If you have Java installed it should associate itself with jar files and run when you double-click them.

Also I found bugs, such as the program not writing to where you say to. Here's a fixed version:
Download v0.3

I find Jar's load quicker when launched with batch files anyway.

So i have a .bat file with the code "@java -jar pcutmp3gui-0.2.jar" and pcutmp3gui-0.3.jar in the same folder. I double clicked the bat file and nothing happen.

Can we please get a noobie guide or the gui with an exe

Thanks

Using a batch file doesn't change that it's still a command line tool. Therefore it just closes immediately after being started if supplied with no further arguments. You need to start it from a command prompt that's already open (e.g. starting "cmd.exe" from the "Run..." menu item in the start menu).

I honestly have no clue when it comes to these things. I'm just gonna wait til someone can get this on exe

Not so - the batch file, when double-clicked, should launch the GUI, sit there blank, and then close when the GUI is closed.

Try changing your bat file to:


.. and see what is returned. I just had to upgrade from 1.5.0_06 to 1.6.0 to get the GUI to run (pcutmp3 worked fine with 1.5).

NB: you can type (or add it as a line in the batch file):


... at the command promt to see what version you have installed - or open the Java item in your Control Panel. You can update from the Control Panel by the looks of it.