about clipping

2001-10-12 15:07:27

I'm newbie, but there is something about clipping I wanted to say. When you encode a file, sometimes mppenc tells that there were x internal clippings (or something similar). I use a batch file to encode all the songs on a directori with insane profile, and I made it the way it told me wich of the files (if it was the case) had were clipped. Now, if a file isn't clipped, I don't have to have problems at decoding in anyway and so, which is the problem with clipped files if the encoder tells you when encode a file if it is clipped or not? On the other hand, if reencode a file because last time encoder told you had clipping with reescale option, when I burn on a cd, will have some tracks different volumes compared with the original cd?. Well, that was all I wanted to say, thanks (and sorry for my english)

about clipping

Reply #1 – 2001-10-12 16:14:55

I'm not quite sure what you mean with the first part, but yes, if you use -scale when you encode a track, it will slightly alter the volume and it will stay that way even if you decode it and burn it to a CD.

about clipping

Reply #2 – 2001-10-12 18:45:39

Ok... the fact is that you should use a 0.92 when scaling... 0.98 is not enough in all cases... or if you have the time:

Sound Forge the Wavs:

1) Clipped Peak Restoration (-6db)
2) Vinyl Restoration (someone will kill me for this - but restores the dynamics very well) with NR & Crackle disabled.

3) Normalize to 99%.

4) Put a little NR there if you think it has too much hiss.
Still use the -scale in mpc or lame anyway.

The worst thing you can get is people telling you that the mpc/mp3 sounds better than the CD.

about clipping

Reply #3 – 2001-10-12 21:10:28

If you're not worried about some other issues with the MAD decoder, you could use the MAD decoder plugin for Winamp which has an automatic clipping detection and attenuation feature.

about clipping

Reply #4 – 2001-10-12 22:07:22

>>>'When you encode a file, sometimes mppenc tells that there were x internal clippings (or something similar).'<<<

hmmm.. i assume you mean x number of clipped samples. am i right?

>>>'Now, if a file isn't clipped, I don't have to have problems at decoding in anyway and so, which is the problem with clipped files if the encoder tells you when encode a file if it is clipped or not?'<<<

if the mp3 file was encoded with enough headroom, then most accurate decoders won't clip the signal. mad for winamp however uses a dither algorithm that may inadvertently cause additional clipping even if other mp3 decoders won't clip at all. this is why the clipping attenuation selector in mad for winamp is quite useful, though not completely effective

>>>'if reencode a file because last time encoder told you had clipping with reescale option, when I burn on a cd, will have some tracks different volumes compared with the original cd?'<<<

yes, the perceived loudness will be different, depending on how u use --scale in lame. if you're using any presets in lame, especially --r3mix, feel free to go below --scale .98 if the levels of the original are high. for most types of "loud" or really compressed music nowadays i suggest that you use --scale .90. sometimes you may have to go even below that in order to have no clipped samples

>>>'you could use the MAD decoder plugin for Winamp which has an automatic clipping detection and attenuation feature.'<<<

imho mad for winamp doesn't work very well for this purpose

i decoded plenty of clipped mp3s with mad for winamp and found that even when clipping attenuation was set to max, there were still a lot of clipped samples remaining in the resulting audio

imo this happens because the clipping attenuation values in mad for winamp are in fact presets of some sort (eg. 1 = attenuate by -.5dB, 2 = attenuate by -1dB, 7 (maximum) = attenuate by -2dB, etc.)

if for example more than -2dB attenuation is required, you will still end up with a few clipped samples here and there, depending on the overload. using the command line version of mad and inputing a value of say -3dB attenuation solves this in producing no clipping whatsoever. additionally, with the command line version you can also disable the dither algorithm mad uses so that no additional clipping is introduced

but don't take my word for it- to find out how clipping attenuation works in mad for winamp, the best thing to do is contact rob leslie

about clipping

Reply #5 – 2001-10-13 00:36:49

Wasn't he talking about mpc!!?

about clipping

Reply #6 – 2001-10-13 00:59:49

oohhh.. 
well, in that case it still doesn't matter as it works similarly to mp3 in that respect. if the peak levels of the original are too high it will likely clip once compressed. so obviously the best way to make sure there's no clipping is to reduce the levels of the original to make some headroom