• Welcome to the all-new HUG. All content has been converted from the old system, and over the next days we will re-style HUG in a more familiar way.

Getting to grips with "clipping" - an ongoing research project

A.S.

Administrator
Staff member
Common experience or not?

Common experience or not?

Firstly - I disagree with Alan as far as his "digital vs analogue sounding" summing up goes - that's not the way I hear it here at all. Secondly - I find myself confused at present and will try to think it through during the course of the day but so far the sources of that confusion seem to be:...To my ears, on my office system [passive pre/English solid state power/1970s-BBC type big box speakers] the passage from uncompressed to compressed bears as as many similarities to a boosted-presence-range to flat-response speaker simulation as it does to a digital to analogue one....
Now its my turn to be surprised! Proof that without a common experience when listening to audio, the very first thing a listener should abandon is (even well intentioned) third party opinion and get down to actually listening and deciding for oneself.

OK, let's consider the DS clip and the two stages of limiting. As a trained listener what hits me immediately the limiting is applied, is the change in perceived bass/mid/top spectral balance in the music. The reality is (we've seen that from the waveform) that those impulsive 'cracks' from the snare drum are first reduced, and then eliminated altogether as the limiting is applied, and then increased. Subjectively then, to the ear (that is, my ear) that limiting action 'warms-up' the sound, it tilts the perceived spectral balance from the top end and towards the bass and middle tones. Yes, the sound is softer and yes, the sound is easier on the ear. That's to be expected: we've seen from the waveform that those fast-rising drum transients of the CD original have been diminished as the limiting has been applied: that's the definition of limiting the amplitude! The key point here is not the limiting per se, that's a technical curiosity, it's how the sound balance can perceptibly change to the human ear when the amplitude of a music signal is altered.

You must be registered for see images attach


Considering what is going on in the ear, when sound limiting is applied, the ear is not having to cope with the impulsive shock wave that is passing down the ear canal, through the middle ear and then rippling along the fluid in the basilar membrane of the inner ear. Physiologically them, the limited sound wave is (in common language) doing the ear a favour by reducing its workload. No wonder it creates a 'softer' experience; there is likely to be significantly less mental processing when there is less energy in the sound wave.

Can we find any agreement that the limited examples are, at the very least, softer on the ear and somewhat warmer in sound? If that's not your experience, that's fine. In the absence of feedback, we're working flat out trying to double guess the reader's experience.
 
Attachments only viewable to members

EricW

Active member
Warm but not good

Warm but not good

Can we find any agreement that the limited examples are, at the very least, softer on the ear and somewhat warmer in sound? If that's not your experience, that's fine. In the absence of feedback, we're working flat out trying to double guess the reader's experience.
Agreed on the softer, agreed on the warmer. As long as I separate those descriptors from any sense of the quality of the sound. That is, those words may be reasonably accurate, but that doesn't mean that the sound they describe (relative to the other) is good. (That alone is something of a revelation.)
 

A.S.

Administrator
Staff member
Dynamics and quality

Dynamics and quality

Agreed on the softer, agreed on the warmer... (That alone is something of a revelation.)
Now this becomes ever more fascinating!

Nobody said that 'goodness' has any automatic correlation with subjective concepts like 'warmth' or 'brightness'.

If you turn up the bass control on your amp, you have a warmer sound, but the 'goodness' of the sound is the same: maybe better. Better if a) you prefer it and consider it to be more natural/lifelike/engaging and b) maybe the demonstrably extra bass out from the amp is compensating for some deficiency elsewhere and the result (technically) is not excessive bass, but now correct bass.

The acid test must be this: if the application of a little level adjustment (i.e. tone controls) at the mixing console or the home hi-fi rig can counterbalance some linear (i.e. frequency response deficiency) in the sound, then that is a one issue: one user's application of tone correction can undo the previous stage*. However, if the dynamics of the sound and the relationship between the instruments has been destroyed, by limiting or compression, that is a non-linear change, destructive, irreversible, once and for all reworking of the music.

The linear change must not be confused with the non-linear, destructive change.

*Example. There are three different stages a commercial recording can pass through: engineers A (recording/mixing), B (mastering) and C (home user with amp/tone controls). If A turns up the bass in a broad sweep across the bottom end to warm up the sound (because he likes it), B can apply the exact inverse cut, turning down the bass in a mirror broad sweep to make it less rich, and C at home can again turn up the bass to suit. Listening at home, there is no degradation in the quality of the sound, merely the quantity. Linear changes are reversible. One man's treble-up can be perfectly annulled by another man's treble down: and be back to neutral again. But if A or B crushes the dynamic range of the recording, no matter what C does or how much he spends, that is a destructive change that cannot ever be reversed. That's why clipping/limiting is so important. It is not undoable.
 

weaver

New member
Warmer means inviting?

Warmer means inviting?

Realise that you may still be typing Alan, but:

'softer/easier' on the ear I'm happy to go with

'warmer' I find difficult as to me that also suggests a richer, more full bodied sound - yes the tilt may be towards the bass and mid, but the richness of that bass/mid also seems diminished - it is warmer but low-fat, to me warmer is by-definition full-fat.

As a strict definition it may be that you look at/hear a bass/mid bias and think 'warmer' - I'm probably more visual than that so warmth also means lush and inviting - even when applied to sound.
 

gakyle

New member
....at the very least, softer on the ear and somewhat warmer in sound.

....at the very least, softer on the ear and somewhat warmer in sound.

The difference may seem small but the CD has a clarity that seems preferable. If I used the "Tilt" control to achieve an approximation of what the compressed sound is like, I would immediately return to a flat setting to listen to the music; using not over-bright speakers in a well damped living room.
 

weaver

New member
Spectral plot?

Spectral plot?

Would it be possible to show a spectral plot of the three clips at this stage please?

[I hope I'm asking for the right thing: a plot of the amplitude relative to frequency for the whole of the clip - I'm assuming that this would show how the balance top to bottom changes as level limiting is applied.]
 

A.S.

Administrator
Staff member
A tipping towards a analogue

A tipping towards a analogue

The difference may seem small but the CD has a clarity that seems preferable...
The CD excerpt is not only preferable, it is what the performance must sound like. It is the reference, like it or not.

The point I was hoping to make (but failed miserably!) is that we can comprehend why some listeners really are attracted to that softer, warmer sound, which is, generally speaking what is referred to in the audio media as 'the analogue sound'. That is what they say: they never mix the words analogue and high resolution or analogue and astonishingly low noise floor, or analogue and brightness or analogue and low distortion or analogue and amazing channel separation or analogue and resistance from physical damage do they. No: they say 'analogue is a warm sound'.

What they really mean, those clips demonstrate, is that when there is less dynamic range (inevitable with analogue, all analogue, ever) there will be a tendency, all other factors remaining constant, of a subjective tipping of sound balance from the HF to the bass/mid. That makes the sound perceptibly 'warmer'. And that's the long and short of it.

Important to remember: we have taken a digital CD source and made it sound warm and analogue like by destructively limiting its dynamic range. However, no matter how we try, we cannot take any analogue recording and make it sound like a digital one. We cannot create dynamic range out of nothing, but we can synthesise the analogue sound from a digital one with impunity.
 

A.S.

Administrator
Staff member
Impuslive sounds in music ....

Impuslive sounds in music ....

Would it be possible to show a spectral plot of the three clips at this stage please?

[I hope I'm asking for the right thing: a plot of the amplitude relative to frequency for the whole of the clip - I'm assuming that this would show how the balance top to bottom changes as level limiting is applied.]
But with respect, you already have that information in the wave plot.

You know that an impulse, a click, a transient, a snare drum, must have lots of high frequency energy or it won't be a transient at all. And that HF energy means sound leaps up in level in the briefest of time, a vertical spike. A note from an electric guitar can't do that. Nor can a violin. Nor human voice. So when you look at what I have marked as the snare drum vertical streaks, that can only mean one thing: lots and lots of (fast rising) energy. It could, of course, be the sound of gunfire: that has a similar impulsive crack as the gunpowder explodes.

The only instrument capable of generating such a fast increase in loudness in such a short period of time is something that is being struck ... so it must be something to do with the drum kit, as we can hear from the correlation of the visual streak and the music itself (limited by the screen cam synchronisation). Just look how the level meters leap when the snare is hit! And it's the quantity and hence perceived quality of the snare that changes so much as the limiting clamps it down in the two middle excerpts. When the limiting is turned off, the snare sings again and you can hear 'air' (i.e. the recording studio) around the impulsive snare crack it as the energy bounces off the studio wall and on to the mic following the note.

No?
 

weaver

New member
Energy from where

Energy from where

OK, got that - but that brings me back to something I was saying in #60, which is that you tell us fairly frequently that very little of the overall energy in a piece of music is in the upper frequencies (most recently I'd guess this was to do with tweeters and how much they actually contribute).

Putting those two things together now, my conclusion would be that an impulsive transient has a large amount of energy but for a very brief time, so it's overall contribution is small but it's instantaneous contribution may be very large.

My previous conclusion was that it was the low frequency element of the snare strike which was producing the spike (as I'd assumed the higher frequencies were low energy) this appears to have been incorrect.
 

A.S.

Administrator
Staff member
OK, got that - but that brings me back to something I was saying in #60, which is that you tell us fairly frequently that very little of the overall energy in a piece of music is in the upper frequencies (most recently I'd guess this was to do with tweeters and how much they actually contribute). Putting those two things together now, my conclusion would be that an impulsive transient has a large amount of energy but for a very brief time, so it's overall contribution is small but it's instantaneous contribution may be very large. My previous conclusion was that it was the low frequency element of the snare strike which was producing the spike (as I'd assumed the higher frequencies were low energy) this appears to have been incorrect.
Yes, the total energy in the high frequencies as a proportion of the entire energy of the music piece is very low indeed. But that is not to say that it is unimportant for fidelity. The higher frequencies are, generally speaking, harmonics of normal (non percussive) instruments like violins, guitars, brass etc.

In the case of the drum kit, there are two factors at play. The 'weight' in the note is from the sheer physical exertion of the drummer. We know from Mark's explanation earlier, that the RMS value of a sine wave is equivalent to it's heating factor, how much it could raise the temperature of whatever it is passing through. Judging from the drummers copious sweating, it's clear that he is transferring his energy through muscles into an impact with the drums, and that defines the 'body' of the note. But the 'crack' of the note is wholly the high frequency content as air rushes away from the point of impact with a particular tension in the skin. Consider the xylophone: also a percussion instrument, but nobody would ever confuse the warm, high frequency less sound of the xylophone with the sharp, bright sound of the snare.

It's the acceleration in the microphone diaphragm which is what drives those streaks in the sound waveform and that fast rise time can only come from some sound source that is accelerating fast .... such as a whack on a snare drum.

Proof? Easy: I've generated some very short, sharp digital clicks. See how they take the same general visual appearance as those snare impulses and you can imagine with some tone shaping how they could be made to sound like a percussive musical instrument.

You must be registered for see images attach


 
Attachments only viewable to members

EricW

Active member
Does "warm" = "good"?

Does "warm" = "good"?

What they really mean, those clips demonstrate, is that when there is less dynamic range (inevitable with analogue, all analogue, ever) there will be a tendency, all other factors remaining constant, of a subjective tipping of sound balance from the HF to the bass/mid. That makes the sound perceptibly 'warmer'. And that's the long and short of it.
That's a very helpful and succinct summary. Thank you.

What this has clarified for me, however, is that there's really no qualitative aspect to the descriptor "warm". For better or worse, I do think "warm" is often equated to "good."

However, what kept going through my mind was rather the terms "alive" and "dead". The original had life (an extraordinarily vague term, I know). By that I mean it sounded to my ears a lot more like what an actual drummer hitting an actual snare drum in an actual room would sound like. To use another vague term, it sounded real (or more real, anyway).

The limited version just sounded dead. It was hard to feel the same sense of involvement. It was as if everything was happening much farther away. If I'd heard it on the radio, listening casually, I might not have noticed anything untoward. But hearing it side by side with the original, the contrast was truly striking (pun intended).
 

A.S.

Administrator
Staff member
That's a very helpful and succinct summary. Thank you.

What this has clarified for me, however, is that there's really no qualitative aspect to the descriptor "warm". For better or worse, I do think "warm" is often equated to "good."

However, what kept going through my mind was rather the terms "alive" and "dead". The original had life (an extraordinarily vague term, I know). By that I mean it sounded to my ears a lot more like what an actual drummer hitting an actual snare drum in an actual room would sound like. To use another vague term, it sounded real (or more real, anyway).

The limited version just sounded dead. It was hard to feel the same sense of involvement. It was as if everything was happening much farther away. If I'd heard it on the radio, listening casually, I might not have noticed anything untoward. But hearing it side by side with the original, the contrast was truly striking (pun intended).
It was, as you correctly say, 'dead' for one very good reason.

There was diminished reverberation.

Why?

Because reverberation (in the studio or hall) is readily excited by impulsive, sharp sounds with lots of HF energy. That energy works its way to our ears (and mic) and tells us how big and airy the space is around the impulsive sound. In other words, from childhood we have learned the relationship between physical space dimensions and echoes, and we draw on that look-up table when assessing a recording. The BBC seem to almost always get that recorded balance right, rather 'wetter' than commercial CDs. Brass and percussion have just the sort of HF harmonic content to blast energy into the recording space, and if the space is not too dead (and the design of a hall/studio must not be too dead), just the right proportion will come to the listener.

When there is a sense of space, there is the possibility of great (classical) sound. If, due to compression, the high frequencies are crushed, the sense of air and space diminishes to our ear/brain.
 

A.S.

Administrator
Staff member
Interpreting sound waves

Interpreting sound waves

OK, I'm back from the gym. In what I can only describe as a near-death experience (the white lights, the tunnel, the eureka moment, my wicked life flashing before me...) in the spin bike group being screamed at by the super-fit instructor, it occurred to me that Mark and I have assumed (foolishly) that readers here can interpret the various pictures of sine and sound waves we've presented. I'll leave Mark to do the really technical stuff with sine waves and so on (he's properly qualified to do so) and address my question to the viewers of my screen cams of sound waves of music.

Tell me honestly: do these audio software screen cams make sense? Can you pick out features in the waveform which could in any way relate to your listening experience, and the general thrust of this thread? Never, ever be ashamed to admit a limit of comprehension here on HUG. It's our problem if you don't understand something, never yours. Just give us the opportunity to try and bring us all up to speed. That's why we need feedback please, otherwise we all stumble along in avoidable ignorance. And there is far, far too much of that in "high-end audio".
 

npoguy

New member
I listen differently now ...

I listen differently now ...

I have to admit that I've been following this discussion since the original thread, but not completely following the technical discussion. However, the give and take between the two of you has started to sink in, so I truly appreciate the willingness to make the info accessible to all levels.

In fact, this thread has already changed the way I listen to everything from my hi-fi to the radio! It's also made me question my assumptions about the "reference" CDs I've relied upon to test new equipment. I know you might not be getting the number of replies you've been hoping for, but please know that people are reading and enjoying!
 

weaver

New member
Interpreting visual representations of sound waves

Interpreting visual representations of sound waves

Thanks for pausing to ask the question Alan.

For me there are two separate aspects to the "do they make sense" question.

Firstly do I know what I'm looking at, do I understand amplitude vs time as shown on a graph? the answer to that is yes, the concept of the amplitude of a passage of music varying with time and being shown visually is something I am now comfortable with (thanks in part to HUG).

Secondly do I know how to interpret that graph, can I look at it and immediately think - there's a spike, that must be an impulsive sound, to have a rise time like that it must be high frequency/high energy. To that the answer is no; by looking at it while listening to the track I can associate it with the snare drum strikes but which aspect of the snare drum strikes was not clear to me.

The second aspect obviously requires two things: knowing how to 'read' a visual representation of sound but then having enough background knowledge about how sounds are made to be able make deductions from it.

Coming back to a simple sine wave, we have been looking at the sine wave representation of the voltage which, when fed to a speaker, produces a sound which can then itself be represented by a sine wave. We then need to know that sound is a series of compressions and rarefactions whose amplitudes with respect to time can be represented by a sine wave.
So we actually have two sine wave representations of sound - one of the electrical signal the other of the sound itself - I do need to remind myself which is which!

Looking at the first clipped sine wave example (post 16) - what I hear is two frequencies - but I don't immediately see two frequencies in the display, from memory there is a way of breaking down a composite wave form into its individual frequencies but I'm assuming that for other people looking at a simple waveform like this that that information is apparent without needing to break it down further? (in post 1 of this thread Mark refers to the harmonics generated when a sine wave clips - I don't know whether these would show as separate peaks or would be 'absorbed' into the overall waveform, I'm guessing it would depend on their amplitude.)
 

weaver

New member
Adding sine waves

Adding sine waves

You must be registered for see images attach


This image should illustrate how sine waves are added together.
 
Attachments only viewable to members

A.S.

Administrator
Staff member
'Drying out' a sound

'Drying out' a sound

It was, as you correctly say, 'dead' for one very good reason.

There was diminished reverberation.

Why?

Because reverberation (in the studio or hall) is readily excited by impulsive, sharp sounds with lots of HF energy. That energy works its way to our ears (and mic) and tells us how big and airy the space is around the impulsive sound. In other words, from childhood we have learned the relationship between physical space dimensions and echoes, and we draw on that look-up table when assessing a recording. The BBC seem to almost always get that recorded balance right, rather 'wetter' than commercial CDs. Brass and percussion have just the sort of HF harmonic content to blast energy into the recording space, and if the space is not too dead (and the design of a hall/studio must not be too dead), just the right proportion will come to the listener.

When there is a sense of space, there is the possibility of great (classical) sound. If, due to compression, the high frequencies are crushed, the sense of air and space diminishes to our ear/brain.
Another post here which illuminates the effect of 'drying out' the sound and how it changes our perception of spaciousness around an instrument.
 

EricW

Active member
Voicing speakers and the moders sound?

Voicing speakers and the moders sound?

Another post here which illuminates the effect of 'drying out' the sound and how it changes our perception of spaciousness around an instrument.
Fascinating. A couple of questions/observations, then.

Is it possible that some of the peakiness/treble overemphasis in modern speakers is due not just to the need to grab showroom attention, but also - consciously or unconsciously - as a compensation for the missing "real" treble in modern recordings?

It therefore seems to me that it matters a great deal what a speaker designer listens to in assessing the validity of a design - correct? It seems to me that if a recording has a natural, unclipped, uncompressed treble content, the speaker designed to reproduce that recording will naturally turn out more balanced (if the designer's skills are up to par) that a speaker "voiced" with modern, clipped recordings.
 

jair44

New member
Harbeth Amplifier with built-in Graphic Equaliser?

Harbeth Amplifier with built-in Graphic Equaliser?

Fascinating. A couple of questions/observations, then.

Is it possible that some of the peakiness/treble overemphasis in modern speakers is due not just to the need to grab showroom attention, but also - consciously or unconsciously - as a compensation for the missing "real" treble in modern recordings?

It therefore seems to me that it matters a great deal what a speaker designer listens to in assessing the validity of a design - correct? It seems to me that if a recording has a natural, unclipped, uncompressed treble content, the speaker designed to reproduce that recording will naturally turn out more balanced (if the designer's skills are up to par) that a speaker "voiced" with modern, clipped recordings.
The more I follow this thread, the more it seems that some form of equalising at the user end is essential. Whatever method of listening I choose I prefer the sound to be on the warm side of neutral. Sometimes it can verge on muddy but a strictly neutral balance would leave half of my cds sounding cold and clinical. Try listening to Jazz and Classical on a system with a cold midrange eg cheap car/pc speakers.

Now if it is true that the major criticism of graphic equalisers (that they degraded the sound through contact losses in the switches/processing) is a load codswallop, then why not bring them back?

Perhaps some means of electronic equalising such as that found on cheap MP3 players could be incorporated into the design of future amplifiers. Of course only a company with no interest in bringing out an 'improved' version every 6 months could do this. Perhaps only Harbeth could do this.

That would finally leave us with only 2 things to be concerned about, the recording quality and the loudspeaker.
 

A.S.

Administrator
Staff member
The consumer, the music and the speaker designer

The consumer, the music and the speaker designer

Fascinating. A couple of questions/observations, then.

Is it possible that some of the peakiness/treble overemphasis in modern speakers is due not just to the need to grab showroom attention, but also - consciously or unconsciously - as a compensation for the missing "real" treble in modern recordings?

It therefore seems to me that it matters a great deal what a speaker designer listens to in assessing the validity of a design - correct? It seems to me that if a recording has a natural, unclipped, uncompressed treble content, the speaker designed to reproduce that recording will naturally turn out more balanced (if the designer's skills are up to par) that a speaker "voiced" with modern, clipped recordings.
That's an insight I'd never considered before. New thinking.

If I understand what you are saying, it is this: (correct me if I am wrong...)

'We've seen in this thread (post 55) that limiting (or compressing) the natural dynamics of a sound recording has, as a by product, the effect of making the sound less toppy and subjectively tipped towards the lower registers because the HF content is diminished. Most/all modern popular music recordings are compressed, and this has been reported as The Loudness Wars. It follows then that if a loudspeaker designer targets his product at an audience who listen to those modern recordings with reduced 'air', he may intentionally boost the HF (tweeter) level in his speaker design to introduce some 'brightening' of sound, to partially overcome the losses in compression and restore a more open balance to the listener at home.

Hence, the loudspeaker designer, the loudspeaker sonic balance and the musical diet of the audience are somewhat bound together.

However, if the music replayed over such an optimised loudspeaker design is naturally balanced (i.e. classical/acoustic recordings) the loudspeaker will be exposed as having too much HF energy, and may sound unnaturally bright and fatiguing...'

Is that your argument? If so, that explains much.
 
Top