What does clipping sound like?

R

Romulus

Junior Audioholic
I was listening to sarah brightman's las vegas dvd tonight rather loudly and heard faint popping from the speakers :eek: The sound didn't fall apart so I didn't think I was maxing the speakers out (b&w 684s), but there was audible popping of sorts. Was the what amplifier clipping sounds like?
 
Haoleb

Haoleb

Audioholic Field Marshall
If you have ever heard the red hot chili peppers album "californication" you'll get a pretty good idea of what clipping sounds like. ;)
 
F

fmw

Audioholic Ninja
I was listening to sarah brightman's las vegas dvd tonight rather loudly and heard faint popping from the speakers :eek: The sound didn't fall apart so I didn't think I was maxing the speakers out (b&w 684s), but there was audible popping of sorts. Was the what amplifier clipping sounds like?
In the old days, electric guitar players used to overdrive their amps in order to get a fuzzy, distorted sound from them. That is the sound of clipping. You can hear it on most old blues recordings. These days they reproduce the sound with pedals and digital samples without actually clipping the amps.

What you heard is something else - possibly something in the recording itself.
 
J

Joe Schmoe

Audioholic Ninja
If you have ever heard the red hot chili peppers album "californication" you'll get a pretty good idea of what clipping sounds like. ;)
That is clipping on the master, and not the same thing as an amp clipping. Not sure if they sound similar or not (I have never driven an amp into clipping, and never will.)
 
TLS Guy

TLS Guy

Audioholic Jedi
Digital clipping

It sounds like digital clipping to me. I bet there was sloppy mastering technique in the CD production. There were not watching the bit meter and ran out of bits. It sounds just like you describe. Analog clipping sounds quite different.

You can see a bit meter in this picture. The bit meter is in the bottom right corner of the right screen.

http://mdcarter.smugmug.com/gallery/2424008#127077194-A-LB
 
davidtwotrees

davidtwotrees

Audioholic General
I don't think I'm alone in wondering about clipping and compression and bad recordings. I'm certainly not worried about clipping my amps or speaker setup-all are pretty robust units. Wouldn't digital clipping be part of the badly recorded, compressed music we talk about so much here? I'm not sure compressed recordings are that obvious and I sure wish I could be told/shown a good way to hear what you all are talking about.
 
F

fmw

Audioholic Ninja
I don't think I'm alone in wondering about clipping and compression and bad recordings. I'm certainly not worried about clipping my amps or speaker setup-all are pretty robust units. Wouldn't digital clipping be part of the badly recorded, compressed music we talk about so much here? I'm not sure compressed recordings are that obvious and I sure wish I could be told/shown a good way to hear what you all are talking about.

No they are different things. Clipping in recordings results from failing to capture the complete waveform during recording. In the analog days this was caused by tape saturation. In the digital era it is caused by "running out of bits" as TLS describes it. In the analog days it was a progressive thing. The more voltage you applied to the recording heads the more distortion you created. With digital, it is, well, digital. It's OK until you over do it. Then it goes immediately bad. Digital recording engineers to to be careful not to overdo levels even moreso than they did with analog recording. It isn't so hard to do because digital recording doesn't have to deal with the tape hiss inherent in analog recording. Digital can and does have a larger signal to noise ratio.

Compression is different. Here in the digital age it means removing data from the recording in order to make file sizes smaller. The idea is that they can remove quite a bit of data without affecting the sound meaningfully. And they can, up to a point. But, like anything done to excess, compression can be bad if it is overdone. If you buy MP3 files on line you will mostly likely receive overcompressed files. If you heard the MP3's I make myself you would consider them true to the original.

Clipping in an amplifier is similar to clipping in analog recording. If the amplifer doesn't have enough power to reproduce the signal requested of it at the level requested, it chops off part of the signal and delivers that to the outputs. This results in distortion which can be anywhere from inaudible to a very fuzzy sound depending on how much of the signal is chopped off or clipped. If you look at a clipped sine wave on an oscilloscope, you will see flat tops on the waves with each wave flattened at the same voltage level. This indicates where the amp ran out of "gas."
 
M

MDS

Audioholic Spartan
Digital clipping as being discussed here is N consecutive max value samples in a row and results in a squaring off of the tops of the waveform.

To put the 'running out of bits' analogy in context:
0 dB is the top of the range for digital audio and is the maximum sample value for the bit depth being used (for example, -32768 or +32767 for 16 bit audio).

The process of analog to digital conversion essentially assigns a value between the min and max to represent the amplitude of the audio at each point in time (how many points in time is the sampling frequency, like 44.1 kHz). If the record level is too high, the resulting value would be greater than the max allowed and it therefore gets 'clipped' to the max. Small sections of clipped peaks are generally inaudible but lots of them spaced closely together can result in the grainy type of sound. If the highs are clipped it can sound sibilant (pronounced 'S' sounds) and if the lows are clipped it can sound like a thud.

Dynamic compression that everyone complains about can cause clipping just as clipping can occur during the intial recording with too high of a record level. Dynamic compression is not related to lossy or lossless compression which are data reduction techniques that make the file size smaller (lossy compression discards some of the data while lossless does not).

I'm sure I can dig up a song or two with lots of clipped peaks and post a picture of the squared off waveform peaks.
 
J

Joe Schmoe

Audioholic Ninja
I'm not sure compressed recordings are that obvious and I sure wish I could be told/shown a good way to hear what you all are talking about.
I have heard a few (very few) CDs on which I found compression and/or clipping to be audible. The most obvious aspect is that they are all loud, all the time (nothing quiet for contrast.) Sometimes, they also have a kind of hiss to them (not unlike AM radio.)
 
TLS Guy

TLS Guy

Audioholic Jedi
How good is good enough? Problems galore!

We seem to have a nice discussion here. I'm going to risk setting off further controversy, I hope without causing too much offense.

First of all FMW and MDS are absolutely correct about digital clipping and the differences between analog and digital compression.

Lets go to first principles.

Dynamic range compression means narrowing the decibel gap between the soft and loud passages in the program. On pop radio stations it is carried to the extreme, so that it is all loud. Dynamic range compression can be accomplished by both analog and digital devices. They are called audio compressors.

Now in the digital domain we have another form of compression, which in the lossy forms throws away parts of the program the authors of the codecs in question think we won't notice. Most codecs have degrees of compression that the creator of the file can choose. Compression rates vary from chucking out nearly all the bits to just over half the bits. The purpose is to make the file smaller. If it is audio to be streamed, to reduce the bandwidth.

There are also lossless codes such as FLAC, which is an open codec, Apple lossless and the new Dolby and DTS codecs for HD and Blue Ray video discs among others. These codecs reduce the file size significantly, but not as much as the lossy ones. The difference is that they are code/encode, so when the file is recreated to play back, all the bits are there, and it is CD quality. In my view none of the lossy codecs, even at their highest bit rates achieve anything approaching CD quality. A CD plays back at a bit rate of 1411.2kbits/sec. The maximum bit rate catered by mp3 is 320 kbit/sec and 128 kbit/sec is usually what is on offer! They have the gall to call this near CD quality.

Now a word about dynamic range and signal to noise. The CD has a theoretical range of about 100 db or so. To take care of the full dynamic range takes skill, as digital systems don't clip gracefully but with sudden unpleasant effects, if you really hit the wall, believe me everyone will notice. Why is there a noise floor anyway? In analog systems it is the loudest passages that were always the headache, tape saturation, over modulated record grooves and such. All sandwiched between tape hiss and record groove noise. However in digital systems it is the fade, particularly dying notes, in ambient spaces that are the problem. Digital systems can only right 1 or 0. At a point the program has to choose between 1 or 0, and so there is a signal level were the error becomes 100%! So white noise is added to the program in small quantities. This is called dither. The effect is to allow sounds to fade into the noise floor without having 100% quantitization error.

So what is the point of this discussion? In a word greed. Bandwidth costs money. So in commercial terms, the question becomes how many bits can we throw away before significant numbers of individuals complain.

They get away with it for two reasons. There are few critical listeners anymore. People are listening on the run to ipods via headphones, or low fi docking stations. Everything is pop culture geared. Now here I'm going to get controversial, but I believe audio errors are much less noticeable in popular music than most forms of classical music. An mp3 file at 320kbits/sec is just not adequate when played back through a state of the art reproduction system. I have particular torture tests that I use to evaluate these lossy codecs on my digital audio workstation. On choral music they are particularly dreadful. If it is boys voices, with large organs and or orchestra in ambient cathedral spaces the results are tragic. The bass depth is not there, the ambient envelope is just plain weird, when the choir hits the high descant there is quite often gross twinking, which is really excruciating and lastly there are episodes of stereo collapse.

I was having trouble getting a CD of anthems by William Boyce from Christ Church Oxford. I purchased an mp3 download. It was a complete waste of money. Eventually I found the CD.

I and others have been screaming to the record manufacturers who sell downloads to offer FLAC downloads. Chandos have just added FLAC downloads to their on line catalog, bless them. I hope others will follow. The Philadelphia Orchestra have offered Flac downloads for some time. They are excellent.

What I'm getting at is that we all have to lobby for standards to be maintained and improved, and fight the good enough for most people mentality. The next big frontier is going to be digital radio. I have experience of that in visits to the UK. The BBC Third program has the highest bit rates, however when comparing DAB with FM radio, the FM radio wins on quality by a mile. In the UK analog FM is scheduled for a phase out. British music lovers are fighting a vigorous campaign, to keep FM, or have DAB that is true CD quality. The problem is the broadcasters want put multiple stations on one carrier, and split the bandwidth between them. It's that old greed again.

Everyone stay tuned on this one. Great music and the dedicated musicians who perform it, deserve that we demand, and fight for TRUE CD QUALITY as the MINIMUM standard.
 
M

MDS

Audioholic Spartan
However in digital systems it is the fade, particularly dying notes, in ambient spaces that are the problem. Digital systems can only right 1 or 0. At a point the program has to choose between 1 or 0, and so there is a signal level were the error becomes 100%! So white noise is added to the program in small quantities. This is called dither. The effect is to allow sounds to fade into the noise floor without having 100% quantitization error.
I don't quite agree with the way you worded that although the gist of it is correct. It's not that you can only choose between 0 or 1. Once you have a digital file it is nothing but ones and zeros (actually it's 16/20/24/etc bit numbers).

Quantization error comes in due to the fact that you have integer valued samples. If your bit depth is 16, the range will be -32768 to +32767, but what if the 'real' value should be 32500.5? You have to choose between 32500 and 32501. Dither is added to smooth out that error so that when you 'connect the dots' so to speak on reconstruction you get smooth(er) transitions from one sample to the next. There are many different dithering algorithms.
 
davidtwotrees

davidtwotrees

Audioholic General
TLS Guy, that was an excellent description. Usually you guys get to technical and I hear Charlie Brown's teacher in my head- "wa waaa wawa, wa waaa wa wawawawa."
Although, I must say, we have some recovering audiophiles here, that have dbxed cd vs 320mp3 and could not hear the difference. There is also a thread here with wired science that tackles this very question.
While I don't own a "reference" system, I certainly have a "good" setup, with room treatments as well, and I am hard pressed to tell diffs in my music server's mp3 320 vs the cds played through my Denon 3910. MAYBE, I can hear the high end rolled off a bit, but that may be psycho acoustics.......
As to demanding the highest quality downloads from the industry, you are preaching to the choir, TLS Guy! What few DVD-A's I own sound stunning and I think that should be the standard, but we know how that one went!
 
newsletter

  • RBHsound.com
  • BlueJeansCable.com
  • SVS Sound Subwoofers
  • Experience the Martin Logan Montis
Top