Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[njkyle]

I use MC 10 with an iPod and I have used mp3gain to adjust all of my 3,800 tracks to 89dB (some, such as classical, are album adjusted though) I did this to level the volume while avoiding clipping. But I have a question lurking at the back of my mind...

I have read a large number of posts and I have dipped into some of the technical documentation on MP3gain and ReplayGain, but I am not clear on where the clipping effect actually happens. Is it:

a) a limitation of the encoding in some way, such that sound amplitude in the file is represented by a number that has a maximum;
b) a limitation of the design of decoders
c) a supposed limitiation of players.

That is, if replay gain is non-distructive / reverable, couldn't some high quality players play the sound no matter how high the dBells are?

[GHammer]

Digital clipping is when you run out of bits to represent the sound.
You have 16 bits for example and during the music suddenly there is a pasage that is quite loud and needs 18 bits to represent it.

What happens?
You get a square wave because there are only 16 bits to record info in, you get to that value and hold there until the levels decrease again. Square waves usually are not pleasing to the ear.

So, when you compress music into MP3 format there is information thrown away the decisions on what to lose are made by psychoacoustic calculations.

I have a CD, the CD is quite loud, right up to the '0 db' level if you will.
I rip it to MP3. The psychoacoustic process will increase certain frequencies. Now what was 'on the edge' will be too loud when it is played. It will now need more than 16 bits to represent itself.

My understanding of what 'clipping' means in the MP3Gain usage is simply that the recreated volume will clip if that gain level is used.

Could a player make up for this? I guess it could, but then it would not be recreating what you give it. Could a decoder make up for it? Again, I guess, but every files would have to be 'analyzed' every play.

[Alex B]

I know what clipping means and how it sounds when analog amps clip, but this "digital clipping" has been a bit gray area for me. Even I have tried to understand it by reading articles about it. I would like to know how the digital clipping actually sounds. Does anyone know if there are audio sample files for this somewhere or how to make them by yourself?

One thing I do know is that when digital audio is compressed in the usual lossy encoding formats it's not anymore 16-bit or any other bit depth. It changes to something called floating bit depth.

I have read that in theory it is better to output the floating files at 24-bit or 32-bit (if the software & soundcard support that) and if the files are 16-bit lossless files they should be played as they are. Though, my ears don't hear any difference.

[njkyle]

Thanks for the replies...

If you want to listen to some sample "artifacts" from MP3 encoding, go to:

http://ff123.net/training/training.html

Can't remember if clipping is included.

Generally, the Hydrogenaudio forum:

www.hydrogenaudio.org/forums

 is a pretty good source for this type of info. I should post the question there too.