joining mp3s without re-encoding

There are many results for this on google, but I'm wondering if anyone here has a program of choice for this.

I want to join multiple mp3s (with different bitrates, some CBR, some VBR) into one mp3 without re-encoding. I don't care about the metadata, it can be blank. The program has to be free and for windows (xp or 7).

Any help?

I've used Camtasia Studio 7 to do this for ages now. Not the most efficient option of course, and I don't know for sure if it re-encodes or not, but you can set it to 192kbps output which does the job gooed enough for me.

You could probably Frankenstein something together with a hex editor if you got really familiar with the file format. ;)

...

What? Don't look at me, I have no clue how to do something like that!

I knew that it was possible to split mp3s without re-encoding. Presumably you could stitch different MP3s together without re-encoding if everything else is equal (bitrate, sampling frequency, number of channels, encoding type) but not if any of them are different. At most you might be able to stitch a bunch of CBR and VBR files together (if everything else is equal) in a common VBR file, but a common CBR would require a sophisticated "lossless" MP3 to MP3 transcoder (not re-encoder).

Also, any "stitching" will have to be done on an integer number of MPEG frames, which might introduce some silence or pops at the splices. Editing away silence (eg. to make a megamix) will require re-encoding at least the splicing points.

TimeOfDeath said:

I want to join multiple mp3s (with different bitrates, some CBR, some VBR) into one mp3 without re-encoding.

So you want to take things that are differently encoded, and mash them together into one file with differently encoded bits and pieces?

Good luck playing that back...

I think that practically every MP3 player supports VBR-encoded files (every frame specifies its own bitrate) so unless it's a very old/broken player, it can adapt even to a stream of "patchwork" frames, which is what VBR is.

However to join them without any re-encoding, there are several conditions to be met, which I already outlined, and certain times of splicing might not be possible without at least partial re-encoding.

You could try mp3DirectCut which seems to do what you want (I don't guarantee that any of the downloads will be free from crapware, so proceed at your own risk).

Use Audacity, it’s free (as in free speech AND free beer).

I'm wondering if simply concatenating the files then running mp3packer on them (to regenerate Xing/LAME tags) would work. I'll have to experiment when I get home tonight...

Edit: It works fine, assuming the files have the same sample rate and number of channels. mp3packer will complain that the number of frames differs from what it expected, but it still does the job. I only tested with foobar2000, though...

ducon said:

Use Audacity, it’s free (as in free speech AND free beer).

That it is, but it won't help him with what he's trying to accomplish.

http://mp3wrap.sourceforge.net/

Related to this question (maybe) does anyone know of a piece of software that can analyse all the mp3 tracks in a folder and then set them to the same effective relative "loudness" without re-encoding? We all know that MP3s from different sources can sound quite different loudness-wise. Basically, I want to be able to throw a bunch of MP3s together in a folder and have them play at volumes that make them seem as if they are part of the same album instead of the levels going up and down between songs.

I used to have a program (I forget it's name) that could do the loudness part. It would analyse the files to get a "how it seems to the human ear" kind of value and then set all the tracks to a suitable volume relative to each other. However, it re-encoded the tracks and the loss of quality was noticeable.

Maybe it's something that technically can't be done? I don't know enough about the mp3 format to know if re-encoding would be needed when a change in volume has been made.

[edit] Hmmm... this might do http://mp3gain.sourceforge.net/ [/edit]

http://wiki.hydrogenaud.io/index.php?title=Replaygain

In theory, one could do a lot of "lossless" manipulations (or more correctly: reversible) on MP3s by directly manipulating and transforming the mDCT coefficients, like changing EQ, loudness, volume, and even converting to other compressed formats as directly as possible.

The trouble is, you'll probably not find a software that does that outside of a university's computer lab. The kind of groundwork required for such a lossless manipulation engine is perhaps justified as part of an MSc or even a PhD, but it's too niche for real world usage.

If you don't need to produce a distributable downmix file, but only need this for your own listening, just use a playback-time equaliser.

ReplayGain doesn't touch the audio data. It's just a little tag that a player reads to adjust all that on-the-fly.

It's hardly an obscure thing in a research lab.

ReplayGain does exactly what I described, but in real time, aka it's a playback-only manipulation, and is far from obscure, of course.

But what the OP wanted was a software that can do a similar manipulation to the data itself, and write it back to a (or even the same) MP3 file without actual re-encoding, and that falls straight into black juju territory. MP3gain is the exception that proves the rule, as it actually does what I described.

However it's always better to do such manipulations in a non-permanent way, if it's just for YOUR listening pleasure.

Thanks for the suggestions, I haven't tried them all but mp3wrap was tiny/easy to use and seems to work great for this (merging multiple mp3s submitted by different people into one giant mp3 for uploading to youtube).

cat song1.mp3 >> song2.mp3

This should work?

No, it shouldn’t. mp3 is not just like pure text or pure analogic sound.
Open a mp3 file with an hexadecimal editor like ghex, you will understand.