--xlevel With Mpc? Options

[Rizban76]

Can someone tell me the plusses and minuses of using -xlevel? There seems to be a significant minority of MPC user who rely on this switch. The ducumentation is pretty sparce, something about an alternate clipping prevention scheme.

--Rizban

[westgroveg]

I also would like to know what experienced people like Dibrom, Frank Klemm etc use/recommend. As far as I know the minuses are, –xlevel at times may not remove all internal clipping & future decoder compatibility could be a problem.

[ErikS]

Ehm. It's not a minus that it doesn't remove all internal clippings - you have to compare with the alternative of not using it! You'll get much much less clippings with --xlevel than without. Secondly it is compatible with future decoders - the problem is that it is not compatible with old decoders.

[CiTay]

Okay, here again is a summary i posted in an older thread:

What is the --xlevel switch good for?

> It prevents the internal clipping error in MPC SV7 (current StreamVersion).
> I think, when the codec was designed, Andree Buschmann couldn't imagine
> that one day the CDs would be as heavily overdriven as they are now.


Quoting Frank Klemm:

> This error is a design error of SV7. It was underestimated that
> CDs can be permanently clipped with hundreds of clipped
> samples at the one end jumping directly to some hundred
> samples clipped on the other end. Affected files have a serious
> degeneration of audio quality behind. If you would press vinyl
> with such a level the playtime will be around 8...9 minutes per
> page due to the huge excursions.
> Normally such files should be reconstructed using some audio restauration
> software to reduce distortions below 7%.
> Please also note that most sound cards can't handle full scale
> audio. They distore below 0 dB, you don't need digital overloads.


What happens during internal clipping, quoting Andree Buschmann:

> Let's say the scalefactor range is not enough, then the subband-samples
> will exceed the allowed range and the samples will be hardclipped!
> The decoder cannot reconstruct the original waveform (even if gaining it by -x dB).


What the --xlevel switch does to prevent this, quoting Frank Klemm:

> It uses scale factors outside 0...63, which is mostly possible by
> relative addressing (instead of the 6 bit absolute addressing).
> The first solution [hard-clipping] is compatible with all decoders,
> but generates higher coding errors, the second solution [--xlevel]
> can only be played by Klemm-based decoders (which support scale from -4...127).

My advice is to use --xlevel, the compatibility is really not such a problem. All newer decoders are compatible. The last incompatible decoder is from mid-2001.

[don_pipo_corleon...]

so using --xlevel will prevent coding errors due to clipping in the .mpc file by using a 'smart' scale...
if i understood well, original level is restored after a decoding process...
am i right?

[ErikS]

Sort of, yes. But it's not scaling as you do with the --scale switch and then restoring by "unscaling". It has increased dynamic range (by 3dB I believe)


(Is this correct? Approximately 2 times as many scalefactors -> MaxValue = 2*OldMaxValue -> 3dB increase?)

[don_pipo_corleon...]

questions:
is the original level restored after decoding?
is --xlevel coherent for an entire album? (eg. if i would decode then a burn this album)
the encoder doesn't tell how and whether it deals with the dynamic range...

[ErikS]

is the original level restored after decoding?

Yes!

is --xlevel coherent for an entire album? (eg. if i would decode then a burn this album)

Yes!

[don_pipo_corleon...]

Yes!
Yes!

okay...
sanx for the light!

[Rizban76]

Sounds good. I'll add --xlevel to my command lines. I take it the next version will address this in the default profiles.

Thanks for the info.