I've encoded some of my music to MP2/192k with MPEG DJ Encoder and now would like to recode to MP3. This may sound like a no-brainer for the hi-fi guys out there but can you give me a few pointers on how to minimize the quality loss?

For example, I remember decoding some of the MP2 to WAV for recoding a long time ago but the result was maximized (volume 100%) even tho I have normalized all my music to 98%. Am I getting loss of quality there too? Do I have to worry about the decoder as well as encoder (I'm planning to use LAME.DLL with ABR 192k or 224k).

Altho if anyone knows if there is a MP3 GAIN type of program that can handle MP2 files then this conversion would not be needed.

Thanks for any help!

Normalizing is entirely pointless because it doesn't accomplish anything. As you just experienced, it won't even prevent clipping on decoding.



If you can decode to floating point that might be an option (foobar2000 can do this, as well as handle the MP3 encoding).

Any reason why you want to use ABR?

What is floating point? Does this method preserve the id3 tags? ABR because that is the only variable bitrate that gives me some idea on what the quality and size is compared to CBR with the same value. Are you saying ABR doesn't produce the greatest quality?

It'll preserve the "clipped" samples. Has nothing to do with tags.



I'm not sure I understand what you mean here.



ABR would be optimal if you can do 2-pass encoding. LAME does not have that capability.

VBR should be better otherwhise.

The tags really have to survive the conversion. It's a two day work otherwise. And about the LAME options; I'm using Audiograbber's frontend to do the job (lame_enc.dll) is this ok? I don't have to use the command line version to get the best quality out do I? There are many VBR methods (old/new/mtrh) what should I use and what quality number would be equal to 160-192CBR?

In your case (only 192 kBit/s MP2) I would use -V3 or -V4 presets, because there won't be much content above 16 kHz in your MP2 files.
Did you use stereo or joint stereo with your MP2 encodings?

To keep the tags the best transcoder will be foobar.

Edit: DSPguru once made a program that could change the volume losslessly. This is a bit harder than with MP3, because scalefactors for each subband have to be adjusted.
But I did not test it, maybe someone else can help you there.

Nobody mentioned it, so I guess it should be noted that BeSplit can adjust gain of MP2 files.

A little off topic but here is a LAME 2-pass ABR encoder : http://translate.google.com/translate?u=ht...Flanguage_tools

Yes, and if you Google for mp2gain, it's the second link!

http://besweet.notrace.dk/besplit.html

Hope this helps.

Cheers,
David.

In my experience, using besplit's mp2gain sometimes results in clipping when the normalised file is decoded. From what I remember it's not possible to use besplit to apply arbitrary amounts of gain -- it only does volume maximising.

That is a bug with all BeSplit versions. Unfortunately. You can try to adjust the volume by negative dB values but unfortunately BeSplit ignores negative values (although they should work) and treats them as positive values. For example, for applying -4dB the BeSplit logfile contains a line: " "Asserted -4.0dB to ...." but in fact the gain was increased by +4dB. I found out that there is no work-around, the mp2gain function of BeSplit increases the gain - always. That makes BeSplit almost useless for sound samples if you want to avoid clipping and no bugfix is in sight (BeSplit seems to be unmaintained now).