Seeing as audio has all these clipping problems, why has nobody made a 2-pass mp3 encoder to stop clipping? I know windows media audio 9 has 2-pass VBR audio encoding and probably stops clipping. this would be much easier than having to run mp3gain all the time. I have been trying MusePack, and wow it rocks. Will future versions of musepack not have clipping problems? or Will there be a MPCGain?

Regarding MPC:

1) Use --xlevel for encoding.

2) MPC has replaygain support so I don't see where the problem is.

i am using the recommended mpc settings with xlevel. I read in another thread that xlevel only stops clipping when used with replaygain. Ok i shall try replay gain. sorry im new at this

i dont thinkthat the 2-pass vbr encoding has anythin to do with clipping.
2-pas vbr as just vbr with a giving bitrate as average bitrage.

in other words its when you need a speciel average bitrate but will like the advantage of vbr.

the first pass analyse and the second pass make sjure that the size fits what you need in the end

i see. When i rip my music to MPC with EAC could you set it too automatically run replaygain after encoding?

In theory yes, but in practice no. ReplayGain has to process the entire album to calculate the correct AlbumGain values, this is why you should run it seperately on each album folder.



Well it's more complicated than that... --xlevel works regardless of Replaygain. It's just that --xlevel is not of much use when you don't also use ReplayGain.

Well, in my experience, if when you play the music you attenuate the output (with foobar attenuator or volume control for example) 2 or 3 dB, nearly all clipping goes away, and the little one that remains is inaudible. However, this doesn't eliminate the possibility that there can be some puntual audible clipping in very rare and extreme cases. If you attenuate the output a few more dBs, all possible audible clipping goes away.

--xlevel attempts to eliminate internal clipping by the encoder. Unless an album is extremely loud (Red Hot Chili Peppers: Californication for example) --xlevel is usually successful at this. --xlevel doesn't change the volume.

However there is also decoder clipping. This is what replaygain attempts to fix and is very successful at doing it by lowering the volume to 89dB. But...there are still cases where some music (classical I think is a good example) clips at 89dB.

Replaygain gets rid of almost all decoder clipping, --xlevel gets rid of most internal clipping. If you use just replaygain then the results won't be bad at all, actually quite good other then internal clipping. If you use just --xlevel chances are you will hear a lot of clipping still. Best option is to use both methods.

SV8 will eliminate the need for --xlevel since internal clipping won't be an issue anymore (just like it isn't an issue with MP3 or OGG). However replaygain will still be needed to fix decoder clipping.

Just on a side note... most if not all lossy formats have internal clipping problems.
Mppenc.exe is the only encoder that tells you about that, tho

I have to say I don't know very much about this internal clipping issue. But this is something that sounds strange to me, since, AFAIK all encoders work internally with floating point data, where clipping is not an issue. Could someone enlighten me about the reasons for this internal clipping in MPC and maybe in other compression algorithms too?

Would having a 2 pass encoder not increase quality? It can analyse the whole file and then encode it much better. Like 2 pass divx looks better than 1 pass divx. In 1 pass it does not know about the sounds ahead of where it is encoding, in 2 pass it would.