Jump to content

Talk:Clipping (audio)

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

pops?

[edit]

i have never heard issues with "pops" outside of speaker issues in regards to clipping. there is definately a tonal and temporal distortion issue as new harmonics are added and notes cannot decay in a timely fashion.

"Infinite Peak Clipping"?

[edit]

I recently came across the jargon "Infinite Peak Clipping", which I think means (in the digital audio case) replacing all positive sample values with +1 and all negative values with -1. (Not sure how they handle zero.) Would somebody who understands this term please add it to this article?

MusicScience 17:22, 7 October 2007 (UTC)[reply]

Apparently infinite peak clipping describes a zero crossing detector. ~Kvng (talk) 16:44, 22 April 2020 (UTC)[reply]

Audio examples and description needed

[edit]

There doesn't seem to be an explanation in this article of what clipping actually sounds like. Preferably, there should be an audio example... Esn (talk) 04:22, 5 May 2009 (UTC)[reply]

Please, somebody provide an audio sample to describe better the distortion, if possible. —Preceding unsigned comment added by 87.16.114.239 (talk) 11:10, 12 March 2011 (UTC)[reply]

How about ? ~Kvng (talk) 16:44, 22 April 2020 (UTC)[reply]

Comment on core saturation

[edit]

The article says that -

"A transformer (most commonly used between stages in tube equipment) will clip when its ferromagnetic core becomes electromagnetically saturated."

After reading many articles on transformers, that does not sound quite right. Take an iron core. It's max flux density occurs with no load.

I questioned what causes multi-channel AVRs to clip at relatively low power. Clearly the voltage rails are dropping. But why? The most obvious theory is poor regulation from the power transformer. But what causes poor regulation? My best guess is that the internal resistance of incapable transformers is such that high loads cause more voltage drop in the transformer. In other words, as the resistance of the load becomes closer to the IR of the transformer, more voltage drops across the transformer

It may be beyond the scope of this article to discuss these factors, but I would guess factors related to transformer performance are the main reason for clipping in modern AVR. Nominal rail voltage is likely sufficient. Transformer performance under load is likely the limiting factor, which is one reason why high powered receivers are expensive - they require large transformers.

75.71.201.190 (talk) 22:14, 1 November 2009 (UTC)[reply]

Multichannel amplifiers may clip due to current limiting in the common power supply. Current limiting reduces effective supply voltage. The nature and cause of the limiting depends on the nature of the power supply. ~Kvng (talk) 16:44, 22 April 2020 (UTC)[reply]

Some thoughts

[edit]

It is impractical to discover the transfer function of each individual soft-clipped recording. That would indeed give good reconstruction similar to HDCD peak-extend feature. The tools available in mentioned audio editors can only interpolate everything above a threshold, which works well, but best results are achieved with hard digital clipping where incorrect samples can be unambiguously detected. Any additional process that the signal undergoes makes reconstruction worse, as the "corners" leave their impression upon surrounding audio.

Add "Stereo Tool" to the list of software, in place of the duplicate entry of RX with a version number. It is widely used in radio stations and in the form of Perfect Declipper, but mentioning the full product avoids controversy around the word "perfect".

Cubic spline interpolation is really basic and not used. The author of Stereo Tool has stated that he implemented interpolation using a Fourier Transform with clipped samples ignored, which gives much longer support context than with splines, reconstruct longer segments, and avoid using noise of other nearby clipped intervals in reconstruction. RX employs some learning of tonal components and least-squares interpolation.

If someone wants to hear obvious example of clipping, just listen to "Ultra" by Depeche mode or "The Look of Love" by Diana Krall on any medium other than SACD. -- J7n (talk) 06:15, 29 June 2019 (UTC)[reply]

I have added a Stereo Tool mention. ~Kvng (talk) 13:50, 2 July 2019 (UTC)[reply]

Power supply topologies

[edit]

In the "Causes" section, the effect of filter capacitor choice was described in relation to signal clipping. This is a case of a power supply having a voltage drop in high power outputs. The AC frequency modulation though makes this pretty much a case restricted to unregulated power supplies and high fidelity stereo. A cheap 2000W class d will not have this problem because it will use a SMPS although a cheap 100W stereo might. I made relevant additions to this section, trying not to get too deep into the issue.Nxavar (talk) 14:52, 20 March 2021 (UTC)[reply]

The filter capacitor reduce an AC line voltage component in the clipped signal. The root problem here is the power supply can't deliver the required current. The filter capacitor can only help with that for a few AC line cycles. ~Kvng (talk) 02:38, 25 March 2021 (UTC)[reply]
@Kvng: If the power supply cannot deliver the required current then there will be a voltage drop. My understanding was that this point was about clipping created due to high ripple voltage. A power supply can have a high ripple at high power output but still be within its power rating. This can happen due to the low capacitance of the filter capacitor. In a smps the switching fequency is more dominant in the ripple voltage. In a regulated power supply the ripple voltage is rejected. So the AC line frequency can affect the clipping behaviour only in unregulated linear power supplies. Nxavar (talk) 19:06, 4 April 2021 (UTC)[reply]