Why do mp3/wav cd rips clip without any plugins?

[shredfusion]

When I import songs to start making a mix, they clip slightly without any effects on the track, can someone explain why this happens?

Thanks

[timothyallan]

because you don't have it set to repitch. If you use any other warp mode, it can introduce transients (unless its beats mode and Live is at the exact tempo of the clip)

[shredfusion]

but I do have it set to re-pitch! and warp is turned off so it wouldnt matter

[timothyallan]

Clip the master, or the channel? If it's the channel, and you have it set to 0... I have nfi.

[shredfusion]

it clips both the same amount, and i have both faders set at 0.

for example, i just dragged in a song and the peak level is in the red and reads 0.32

[leedsquietman]

probably because the original file had clipping or intersample peaks that clipped.

Much of today's hyper compressed rock/pop/dance music clips. Not all the time. Some engineers feel that it's better to have the odd clip than flat top a wave through brick walling.

Especially true for mp3 because of the encode/decode algorithm, if you mix music for mp3, you should leave at least 1db of headroom (i.e. turn your track down a tad and set a limiter at -1.0 dbFS, preferably a limiter with OVERSAMPLING to catch any intersample peaks which are clipping but not noticeable)

Also - if you have something with statistical analysis - i.e. Logic, Cubase, Wavelab, Peak, Soundforge - run it through a 'statistics' analysis and see what the peak values are there. I don't trust Live's metering.

[timothyallan]

probably because the original file had clipping or intersample peaks that clipped.

Much of today's hyper compressed rock/pop/dance music clips. Not all the time. Some engineers feel that it's better to have the odd clip than flat top a wave through brick walling.

Especially true for mp3 because of the encode/decode algorithm, if you mix music for mp3, you should leave at least 1db of headroom (i.e. turn your track down a tad and set a limiter at -1.0 dbFS, preferably a limiter with OVERSAMPLING to catch any intersample peaks which are clipping but not noticeable)

Also - if you have something with statistical analysis - i.e. Logic, Cubase, Wavelab, Peak, Soundforge - run it through a 'statistics' analysis and see what the peak values are there. I don't trust Live's metering.

Even if the original clips, it won't go above 0 after being rendered into a stereo file. Intersample peaks? Maaaybe.

[shredfusion]

thanks for all the info, even though im not really sure what any of it means haha. ill have to do some reading. anyway, i ran a statistics analysis with Soundforge, would the "maximum sample value" be the indicator of clipping? and because it says 0.000 does that mean Live's metering is off?

Code: Select all

                              	Left Channel        	Right Channel       
Cursor position (Time)        	00:00:00.000        	00:00:00.000        
Sample value at cursor (dB)   	-Inf.               	-Inf.               
Minimum sample position (Time)	00:00:28.340        	00:00:27.460        
Minimum sample value (dB)     	-0.000              	-0.000              
Maximum sample position (Time)	00:00:31.807        	00:00:30.060        
Maximum sample value (dB)     	-0.000              	-0.000              
RMS level (dB)                	-12.668             	-12.680             
Average value (dB)            	-81.535             	-86.561             
Zero crossings (Hz)           	1,447.77            	1,439.00            

[/size]

[leedsquietman]

run the 'detect clipping' analysis in Soundforge too.

you can be right at the ceiling and not clipping at 0 dbFS, or above or you could still be clipping.

Set the number of consecutive samples at 1 (just for fun and watch those clip markers light up), then do it again at 5. If a sample clips for less than half a dozen concurrent samples, it's not usually audible.

[shredfusion]

wow there was about 200 clips when set to 1. but when i set it to 5 none were detected.

im not too sure what this means.. so the mp3s or wavs from the cd can clip slightly without ever being noticed until you analyze them? cause even the channel meters in Soundforge turn red for clipping. seems weird to me, hard to wrap my head around

[fishmonkey]

basically if it's only a couple of samples in a row that is clipping, it's too short for you to even hear and/or for your audio system to physically reproduce in the first place...

at 44.1kHz you have 44100 samples per second, so each sample is only .023 milliseconds long...

[timothyallan]

I'm going to make a song that is one sample long and include it as a bonus on my next album I think. I will also make it clip.

[shredfusion]

run the 'detect clipping' analysis in Soundforge too.

you can be right at the ceiling and not clipping at 0 dbFS, or above or you could still be clipping.

Set the number of consecutive samples at 1 (just for fun and watch those clip markers light up), then do it again at 5. If a sample clips for less than half a dozen concurrent samples, it's not usually audible.

ok i just analysed some of my own tracks and it detected a lot of inaudible clipping as well.

i dont understand though.. if a signal clips, that part of the original signal was lost, correct? how does Ableton Live know how much it is clipping if that data was lost?

and why does it go above 0 after being rendered? is that because of the intersample peaks?

[shredfusion]

I'm going to make a song that is one sample long and include it as a bonus on my next album I think. I will also make it clip.

id pay 99 cents for that on itunes for sure!

[timothyallan]

I'm going to make a song that is one sample long and include it as a bonus on my next album I think. I will also make it clip.

id pay 99 cents for that on itunes for sure!

In your face John Cage!