Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Question about non-audible clipping (Read 4122 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Question about non-audible clipping

I have an MP3 file (see attachment) with lots of clippings when I look at its waveform, however the clippings are non-audible on my PC with my good headphones.

If I send this file to other people to listen on their other devices (phone, car, etc), will the clippings remain non-audible?

Thanks!

Re: Question about non-audible clipping

Reply #1 –
Try applying a slight negative amplification in a program like Audacity. You will see that the previously clipped peaks are actually not clipped, they were simply floating point values above 0.000, and programs such as Audacity show that as clipping.

Standards-compliant MP3 decoders will handle these values correctly, hence there is no clipping on playback. You don't have to make any changes to the files.

We've covered the same subject in an earlier thread: https://hydrogenaud.io/index.php/topic,113646/topicseen.html

Re: Question about non-audible clipping

Reply #2 –
Quote from: KozmoNaut on
Try applying a slight negative amplification in a program like Audacity. You will see that the previously clipped peaks are actually not clipped, they were simply floating point values above 0.000, and programs such as Audacity show that as clipping.
I learned something today.


OP's file "Before"
OP's file "After" -3dB amplification
On the other hand, sometimes clipping is just clipping.


"Hotel California" from HDTracks after -3dB amplification

Anyway OP, it's nothing worth worrying about.

Re: Question about non-audible clipping

Reply #3 –
Dear KozmoNaut and Apesbrain, I'm afraid I can't agree. Reversible clipping ≠ non-audible clipping. If you boost up the volume of any song way up in Audacity or MP3Gain, you'll definitely hear distortions on playback, although it's reversible.

Did we have any old threads about what makes a non-audible clipping non-audible? The attached song has a peak of 1.39 in ReplayGain and no distortion can be heard, I wonder what makes it so.

Re: Question about non-audible clipping

Reply #4 –
Quote
Dear KozmoNaut and Apesbrain, I'm afraid I can't agree. Reversible clipping ≠ non-audible clipping.
The MP3 is NOT clipped.  It happily goes over 0dB without clipping, and Audacity shows that as potential clipping.

Quote
If you boost up the volume of any song way up in Audacity or MP3Gain, you'll definitely hear distortions on playback, although it's reversible.
If you play a file that goes over 0dB at full-volume into your DAC, your DAC will clip.     And of course when the DAC's analog output is clipped, that's not reversible. 

If you do play it at full-volume into your DAC, the slight clipping caused by MP3 encoding may not be audible.   I've never heard of a case where there were audible MP3 compression artifacts and those artifacts went-away by reducing the volume before MP3 compression.    (I have heard of people "mastering" to -1dB or so to keep the resulting MP3 peaks at 0dB or less.) 

Re: Question about non-audible clipping

Reply #5 –
Quote from: Arrivest on
...what makes a non-audible clipping non-audible?

The attached song has a peak of 1.39 in ReplayGain and no distortion can be heard, I wonder what makes it so.

The audibility of clipping is reduced when it is less frequent and when each individual instance of clipping has a shorter duration. 

If the duration of the clipping is short enough any harmonics created can be moved to frequencies where the ear is less sensitive.

IME most experienced listeners listen for IM as a sign of clipping.  A little clipping of short duration can make the music seem more bright.  Heavy clipping much of the time makes music sound like mud.

This is obviously related to the characteristics of the program material. 

Compressing it can often make it more audibly clipped when it is clipped at all because compression tends to make clipping more frequent and make each instance of clipped samples be longer.

Re: Question about non-audible clipping

Reply #6 –
OK, now I'm really confused.

First of all, let me clarify what I want to achieve: I want to make my file as loud as possible without compressing. So I carefully apply some gain, creating peaks above 0dB here and there, but make sure I don't hear distortions.

Of course, those peaks are not chopped off, they're well preserved, unless I save as 24-bit. I don't know strictly speaking if it's clipping or not, but for the time being let's just call it "reversible clipping".

Now,
Quote from: DVDdoug on
The MP3 is NOT clipped.  It happily goes over 0dB without clipping.
If you play a file that goes over 0dB at full-volume into your DAC, your DAC will clip. 
I admit It makes a lot of sense, and sounds like an epiphany, but it can't be 100% true. If I gain it too much, I can hear distortion no matter how low I turn down the volume. It's easy to prove by doing a quick little test yourself with any file. On the other hand, if I only have small peaks above 0dB here and there, I hear no distortion even if I play it at full-volume.

And to dear Arnold B. Krueger:
Quote from: Arnold B. Krueger on
Compressing it can often make it more audibly clipped when it is clipped at all because compression tends to make clipping more frequent and make each instance of clipped samples be longer.
I assume you meant compressing to a volume that's still above 0dB? Because if you meant compressing to a volume lower than 0, then how could it sound more clipped? It should sound not clipped. (suppose we're still talking about reversible clipping here)

Re: Question about non-audible clipping

Reply #7 –
Quote from: Arrivest on

Quote from: Arnold B. Krueger on
Compressing it can often make it more audibly clipped when it is clipped at all because compression tends to make clipping more frequent and make each instance of clipped samples be longer.
I assume you meant compressing to a volume that's still above 0dB?

Not possible in a digital system.  All signals max out at 0 dB or less.

Quote
Because if you meant compressing to a volume lower than 0, then how could it sound more clipped? It should sound not clipped. (suppose we're still talking about reversible clipping here)

I'm thinking of the following situation:

(1) We have a signal whose maximum value is 0 dB or less.

(2) We compress it and obtain a signal whose maximum value is 0 dB or less.

The way compression usually works is that the signal is processed by some means such as half wave rectification, full wave rectification, or perhaps a true-RMS type nonlinear operation to create a control signal. The control signal is filtered in some simple or complex way to create a more slowly varying control signal. . One reason for doing this is that if the control signal is not filtered, it will result in adding nonlinear distortion to the input signal. The control signal then varies the gain of an amplifier or attenuator. If this is compression, the system is arranged so that as the input signal's level increases, the gain of the amplifier or attenuator decreases, thus decreasing the output signal. In many cases a separate circuit adds steady gain to make up for average losses due to decreasing the signal.

As we compress the signal, short peaks are smooshed so that they are at a high level for longer periods of time. If the peaks are clipped, then the clipping sounds last longer and the signal is clipped a higher percentage of the time.

Often, compression is used to obtain higher average signal levels while smooshing down the peaks.  In that way it can help avoid clipping.

Re: Question about non-audible clipping

Reply #8 –
Quote from: Arnold B. Krueger on
Not possible in a digital system.  All signals max out at 0 dB or less.
You see, I had always been talking about "reversible" digital clipping though. So, by compressing I meant like this:
=>