'Splain it to me Lucy!
- Thread starter drstawl
- Start date
Stealthtech
LOGIC ABUSER
Funny you should bring this up.
I have clipped on tracks before in error, but on playback it sounded fine ???
I have clipped on tracks before in error, but on playback it sounded fine ???
dachay2tnr
One Hit Wonder
Why do you say it's clipped? Wouldn't a waveform that has been "limited" look the same?
Clipped to me means the digital recording exceeded 0db, and therefore the data above 0db is completely missing. If the intital recording was below 0db, but had its gain increased and simulataneously limited (e.g., put through an L2), the waveform would end up flat-topped like that. IOW, the musical data is there, is just isn't being allowed to exceed 0db by the limiter.
Clipped to me means the digital recording exceeded 0db, and therefore the data above 0db is completely missing. If the intital recording was below 0db, but had its gain increased and simulataneously limited (e.g., put through an L2), the waveform would end up flat-topped like that. IOW, the musical data is there, is just isn't being allowed to exceed 0db by the limiter.
Son of Mixerman
Mix Junior
It looks like a soft limit. I'd be interested in seeing what that looked like before because the rise makes it look like a serious reduction in peak amplitude for those samples.
Not all clipping is bad, there are a a few albums done today where the clipping was taseful, loud yet exhibits no obvious distortion. The size of the clip isn't important as the frequencies in which the clip occur. Tree stumps underwater are not always obvious
SoMm
Not all clipping is bad, there are a a few albums done today where the clipping was taseful, loud yet exhibits no obvious distortion. The size of the clip isn't important as the frequencies in which the clip occur. Tree stumps underwater are not always obvious
SoMm
masteringhouse
www.masteringhouse.com
drstawl said:
A limiter is making adjustments to the entire envelope, i.e. lowering and raising volumes (if using some sort of gain makeup) to all of the samples. Clipping a waveform simply truncates any sample that can't be stored for a given wordlength.
To put it another way, clipping is accidental distortion without any regard for the original waveform where limiting is lowering the peak to average ratio to achieve a hotter signal based on the original waveform.
dachay2tnr
One Hit Wonder
OK, let’s try it this way. Let's say I’ve got a piece of music that is 10 samples long. The peak for each sample is as follows:
1 = -8.0 db
2 = -5.0 db
3 = -2.5 db
4 = -1.0 db
5 = -0.5 db
6 = -0.5 db
7 = -1.0 db
8 = -2.5 db
9 = -5.0 db
10= -8.0 db
Now let’s assume I want to add +3.0 db of gain to the entire piece. If I were to simply raise the gain of the entire file by 3.0 db, then samples #3 thru 8 will clip, as they will be driven over 0db, and the part of the sample that exceeds 0 db will be chopped off (i.e., the word length will be truncated as MH said). The result will be digital distortion.
But let’s say that while increasing the gain, I also placed a limiter on the file and set it to 0 db. The resulting output then will be as follows:
1 = -5.0 db
2 = -2.0 db
3 = 0.0 db
4 = 0.0 db
5 = 0.0 db
6 = 0.0 db
7 = 0.0 db
8 = 0.0 db
9 = -2.0 db
10= -5.0 db
Note that samples # 3 – 8 will appear flat, as all have been restricted to 0 db. However, they are NOT clipped. They just have been restricted (limited) by the limiter not to exceed 0 db.
Looked at another way, you really did not add 3 db of gain to all of these samples. In cases where the requested gain would push the sample over 0db, the limiter only allowed enough gain to bring those samples to 0 db (or wherever the upper limit may have been set). Therefore 3 db of gain was added to samples 1,2, 9 and 10, but a lesser amount gain was added to the other samples in order to prevent them from clipping.
Visually, adding 3 db gain across the board versus adding 3 db gain with a 0 db limit, would look the same. Samples 3 to 8 will be flat at 0 db. However only the first case will you actually have clipped samples.
Whew! I have to go rest now.
1 = -8.0 db
2 = -5.0 db
3 = -2.5 db
4 = -1.0 db
5 = -0.5 db
6 = -0.5 db
7 = -1.0 db
8 = -2.5 db
9 = -5.0 db
10= -8.0 db
Now let’s assume I want to add +3.0 db of gain to the entire piece. If I were to simply raise the gain of the entire file by 3.0 db, then samples #3 thru 8 will clip, as they will be driven over 0db, and the part of the sample that exceeds 0 db will be chopped off (i.e., the word length will be truncated as MH said). The result will be digital distortion.
But let’s say that while increasing the gain, I also placed a limiter on the file and set it to 0 db. The resulting output then will be as follows:
1 = -5.0 db
2 = -2.0 db
3 = 0.0 db
4 = 0.0 db
5 = 0.0 db
6 = 0.0 db
7 = 0.0 db
8 = 0.0 db
9 = -2.0 db
10= -5.0 db
Note that samples # 3 – 8 will appear flat, as all have been restricted to 0 db. However, they are NOT clipped. They just have been restricted (limited) by the limiter not to exceed 0 db.
Looked at another way, you really did not add 3 db of gain to all of these samples. In cases where the requested gain would push the sample over 0db, the limiter only allowed enough gain to bring those samples to 0 db (or wherever the upper limit may have been set). Therefore 3 db of gain was added to samples 1,2, 9 and 10, but a lesser amount gain was added to the other samples in order to prevent them from clipping.
Visually, adding 3 db gain across the board versus adding 3 db gain with a 0 db limit, would look the same. Samples 3 to 8 will be flat at 0 db. However only the first case will you actually have clipped samples.
Whew! I have to go rest now.
masteringhouse
www.masteringhouse.com
dachay2tnr said:
Sorry but I have to disagree with this statement. There is not a measurement that allows one to measure above 0 dbFS. As a result, the convention (though can be set via parameters in some devices) is that any 3 consecutive samples at 0dbFS is defined as a clip.
Therefore clipping will result with samples #3-5 and continue to sample 8.
Also the amount of volume reduction is dependent on the ratio and threshold of the limiter. If using a limiter at a 10:1 ratio the volumes will not necessarily be brought up an equal amount. In a brickwall limiter without a softknee (more like the example you are describing) , everything above the threshold will be set to the maximum output volume of the limiter.
dachay2tnr
One Hit Wonder
I was simply trying to express how a waveform could appear flat-topped without actually being clipped. If technically inaccurate in the process, I guess you'll have to shoot me.masteringhouse said:
Change the limit to -0.1 db. The point I was trying to make remains the same.
Also, in this case I was assuming a brick wall limiter, although I probably should have stated that. For my limited mind, I generally consider anything less than brick wall as a compressor (although I know that is technically inaccurate as well).
Anyway, thanks for the clarification.
barefoot
barefootsound.com
The only difference between clipping and limiting is that limiting tends to be more intentional and less drastic.
A limiter essentially clips the waveform, except it can do it any desired amplitude - not just 0dB. However, clipping causes sharp waveform edges that contain a lot of high frequencies. So clipping is often quite noticeable. The "soft clipping" of a limiter subdues the highs by rounding off the edges before flattening out the waveform at the desired level.
Thomas
http://barefootsound.com
A limiter essentially clips the waveform, except it can do it any desired amplitude - not just 0dB. However, clipping causes sharp waveform edges that contain a lot of high frequencies. So clipping is often quite noticeable. The "soft clipping" of a limiter subdues the highs by rounding off the edges before flattening out the waveform at the desired level.
Thomas
http://barefootsound.com