Their expertise lies in making coffins for monkeys
As always, a pleasure to read John Watkinson; his books on digital audio were my bible for years.
Thanks, John, keep up the good work!
Today’s loudspeakers are nowhere near as good as they could be, due in no small measure to the presence of "traditional" audiophile products. In the future, loudspeakers will increasingly communicate via digital wireless links and will contain digital processing. Indeed, the link between IT and loudspeakers is destined to grow …
As always, a pleasure to read John Watkinson; his books on digital audio were my bible for years.
Thanks, John, keep up the good work!
But no progress can be made when science is replaced by bizarre belief structures and marketing fluff, leading to a decades-long stagnation of the audiophile domain.
No. The ludicrous "audiophile" market, thought annoying, is tiny, and should not prove a barrier to any serious new technology. It certainly didn't slow down the adoption of CDs, for example. And thankfully it has no influence over commercial audio. There are no oxygen-free cables in Sony studios.
I am surprised that commentators continue to talk about MP3s and compressing codecs. Storage capacities have already made music compression fairly redundant. A micro SD card can happily house a major CD collection uncompressed these days.
This article covers a lot of ground in a short space, and is therefore superficial in parts, annoyingly so. The author clearly believes that time delay is the overarching consideration in pretty much all audio design, even dismissing (he calls it "debunking") transmission line speakers in a couple of sentences. But this doesn't stand up. As Douglas Self points out in parts 6 and 9 (also numbered 11) of this article
...phase may be important for a base drum, but what about an electric guitar, where even the live performance is provisioned through "legacy" loudspeakers ?
I would add that that live music does not originate at a point source. In an auditorium, the audience members are in widely different positions relative to each instrument, with corresponding wide variations in time delay and sound direction and distance. So should the CD be based on the time delays at seat 34-A or those present at seat 6-C ?
Commercial music passes down an extremely long channel before it reaches your hi-fi. Perhaps hundreds of amplifiers, CPUs, mixing units, filters, DSPs. How likely is it that group/phase delay passes through unchanged at all these points ? Only if all these components have zero group/phase delay is it worth redesigning your loudspeaker crossover unit, and even them only if you think that the effect is gross enough to be audible.
My reaction to what you say is to suggest that what matters here is the distortion introduced by the loudspeaker. We can throw away frequency information—with my ears you can throw away a lot—but there are other elements in the signal, and the designers don't look at the whole picture. Loudspeakers are the place with the big distortions, and the fixes are the low-hanging fruit of audio technology.
My late father, who had terrible hearing, did get something from digital surround sound. It helped him distinguish the background sound from the speech. And sometimes the sound of footsteps had a direction which mattered, and which he could hear.
That just needed an amplifier and a bunch of ordinary speakers. but it was timing information that plain old stereo systems lose.
I knew there was a reason One Direction sound so awful...
No, I think you're jumping to the wrong conclusion about what you think the author believes.
He clearly is focussing on time delay has being a very important consideration, but that does not mean he believes that nothing else is important.
He does mention in the article about how the brain processes frequencies after the brain processes the transient information.
There is no doubt (without doing any research into his background), he's an electronics engineer, and being a fellow of the audio engineering society, I am highly confident he does understand about frequency spectrums, the representation of signals in both time and frequency domains and the transformation from one to the other.
I think he is focussing on one key aspect of audio design which is vitally important which he perceives (and possibly rightly so) is present in amplifier, filter design but which is missing from the last element of the chain in speaker design.
But this doesn't mean he doesn't understand that frequency response isn't important. I'm highly confident he does know it is.
>How likely is it that group/phase delay passes through unchanged at all these points ? Only if all >these components have zero group/phase delay is it worth redesigning your loudspeaker crossover >unit, and even them only if you think that the effect is gross enough to be audible.
First point: manufacturers of equipment (and certainly output stages of DACs) go to great lengths to design an active filter of high order, which buggers up the phase of the frequencies, and correct those phase errors using other types of active filter circuits. I know, because I've seen the circuit diagrams and spoke to the analogue design engineers undertaking the design.
Second point: You're making the assertion that because a system may not be of zero group delay up to the speakers that you don't need to design the speaker elements for minimal group delay. It's a flawed way of thinking. What matters is the total group delay of the entire system and if minimising it only in the speakers helps reduce the total delay for the system, then that's a good thing.
And by the way, from ex-recording turned electronic audio design engineers with whom I have worked, the author is entirely correct in his analysis.
Dave, you are correct.
Audio design engineers with whom I have worked, very much do concentrate on timing, as well as frequency response. But they were not desigining speakers, they were designing mixing consoles with all the analogue (and digital) electronics in that, from active filtering using literally hundreds of operational amplifiers and many different types of active filtering circuits.
I haven't worked for any speaker manfacturers, so I can only assume the author is right when he says speaker manufacturers are not focussing on the timing.
But from discussions I had with colleagues (design engineers and recording studio engineers), there is something missing from all the specifications of audio systems, two systems can look the same on paper, in frequency response, in phase response, but yet still sound different. Generally speaking, our system of measurement, what we are measuring is incomplete.
All very lovely, scientific-sounding and possibly even correct.
However, it all falls by the wayside when faced with the modern fashion of compressing the dynamic range of the music source material to increase its "loudness" to make it stand out from the crowd (or now: blend is and not get left behind) when broadcast or streamed to the user.
So given that fundamental quality limitation, not to mention all the background noise in out lives, the quest for "perfection" whether in loudspeakers, amplifiers or dragging a piece of crystal across some plastic is largely futile.
As it is, for most audiophiles that I have met, the goal is not perfect audio quality. The real goal is to impress their other audiophile friends with the size and cost of their stack.
Sure, garbage in, garbage out - but why settle for garbage? Choose better source material.
Classical music (eg) tends not to fall victim to overcompression - it's wide dynamic range really needs a decent amp and speakers.
And any "new" speakers will immediately have magic woo applied to them. They will be made to look like space ships by designers who are considered more important then the engineers by the marketing department.
You just going to end up "needing" a thousand dollar power cord for each speaker instead of just one for your power amp.
Indeed, one of the benefits of the CD (and therefore digital) revolution in music is that albums were remastered without (or at least with less) compression.
But it also depends on the artist. If you create your music using digital instruments with deliberately limited timbre, then all the speaker improvements in the world won't help make the music sound better than it would with earbuds.
Most modern music is just badly produced almost from day one. That's why I gave up buying CDs and ripping to a lossless format(*). It's just less hassle to download the MP3. My ageing ears help ensure that I don't hear the damage rendered by the studio 'engineers'. In addition most of my listening these days is in the car or via bluetooth headphones on the train or walking around.
I have a reasonable enough system at home but generally don't have the time to just sit and listen. It doesn't bother me too much. I enjoy listening to it - ignorance is bliss and all that :)
(*)WMA if you must know :)
@ Frankee Llonnygog and all
Isabella Stewart Gardner Museum Music Library
Fortnightly podcast of live classical (chamber) music. Mp3 128kB/s bitrate. Very listenable and creative commons licence. Sound a lot better than the bitrate (if you see what I mean)
Enjoy. Pass on to others.
Agree entirely, especially where radio is concerned. The compression and jiggery-pokery that is applied these days sounds almost painful to my ears. DAB was supposed to be the be-all and end-all with the promise of "near-CD quality" and user-adjustable compression (THAT fell by the wayside), and what do we get? More and more stations shoe-horned in at ever-decreasing bit-rates, hideous audio processing to boot (with Radio 4 Extra in glorious MONO, for pity's sake!) with the resultant sound as flat as the proverbial pancake. Why stations also find the need to compress their satellite feeds is beyond me, although they seem to be the best quality available at the moment (for what it's worth). I complained bitterly to to the late Radio Authority about excessive compression and processing some years ago and got back some waffle about catering for the majority of people who would be listening on portable radios or in the car. Poppycock, I say! I have some very old analogue reel-to-reel recordings of good old Alan Freeman doing "Pick of the Pops" on a Sunday afternoon, pre Optimod (or what ever they use now) and and they sound fine. Likewise, very early Capital Radio stuff. Put current offerings into an audio editing program like Audacity and all you get from the waveform is virtually a straight line! No dynamics at all. Absolutely appalling sound. So-called "re-mastered" re-issues seem to have suffered the same fate of heavy-handed compression. Nothing like the originals. Why should classical music be singled out for the non-compression treatment? This all makes a mockery of using high-end audio equipment if the input has been trashed to start off with.
Isabella Stewart Gardner Museum Music Library
Thank you. The internet has been useful today.
vinyl? RIAA curve in/out. digital? hacked and (occasionally) regenerated.
this on top of the compression/expansion, noise gates, preamp distortion, microphone effects, and multistage processing loss of definition all throughout the recording and duplication chain. tape has a 50 dB range with nonlinearities on the low end, and processing like dbx adds additional artifacts. digital recording is itself a series of compromises.
so to start with, there is no "true" fidelity in a commercial source.
I have always had my doubts about the 30s and 40s RCA research that "7% distortion is the point at which the human ear detects." that's as good as they could measure pure tones. and every improvement such as the feedback loop, beam-power tetrode, and differential amplifier stages iimproves on that.
in the beginning and the end, we have analog. careful use of analog technology across the chain provides fewer machine artifacts and a more realistic experience.
I'm smelling more ways to try and push 1-inch transducers and chips and calling that excellent.
when it's just another ploy to get me to dump all my stuff and buy new.
the push side of the market is looking desperately for another gimmick, while the pull side is way tired of "it's all crap, but this is better." we don't want to stuff the landfills and spend all over again out here.
"with Radio 4 Extra in glorious MONO"
Well a lot of the stuff on Radio 4 Extra was actually made in mono. (Hancock, Navy Lark, Much Binding in the Marsh, all the varieties of Round the Horne, Take If From Here) . and some of that has been reclaimed from off-air recording, so accurate rendition is not the prime importance
Mines the one with "Everybody down" written on the back.
The Cramps never did classical.
I like the Cramps.
Therefore, Beethoven can roll over in his grave...
i agaree with the thrust of y0ur comment , but lets not be so eurocentric, all of the acoustical musics, bluegrass, jazz,a Capella, roots etc need wide range reproduction equipment
I listen to the Cramps and Beethoven (though not at the same time). A little John Cage goes with anything though
Thanks for the "Isabella Stewart Gardner Museum Music Library". Just made my morning better.
I think there is a fundamental misunderstanding regarding compression.
An old friend and Grammy winning audio engineer, once told me that "Compression is like sex, it is absolutely necessary for reproduction, can be quite pleasurable between consenting adults, but will get you in to serious trouble if used inappropriately"
"A little John Cage goes with anything though"
His 4'33" reproduces perfectly on any equipment. :-) It doesn't even need to be switched on...
lol! But it depends on the noise floor of the system :)
Abso-bloody-lutely. It's high time that some properly clever disruptive technology exposed the audio emperor's new clothes.
Because it is either:
1. synthesised using electronic instruments
2. subjected to autotune for vocals
3. deliberately manipulated for effect (many modern pop vocals sound like they're been stretched to make them extra trebly and reedy)
4. subject to massive dynamic compression to make it sound louder
5. mucked about with on a mixing desk, overdubbed, artificial chorus effects added etc
Only some carefully-recorded classical music on CD (not radio) might benefit from real high fidelity. Or maybe live pop, but in most cases the performers aren't all that hot without the studio trickery. For everything else, if it sounds good, it is good.
"You see, Piet, I can call you Piet, right? ... Right, anyway you see Mr Mondriaan, we sort of smoothed out the lines on those prints, and we changed the red for blue, because, you know, it was cheaper, and this modern art, it's all imagination anyway, and you know, I think it looks pretty darned good like this..."
If 90% of modern music is indeed just sythetic sound manipulated into a dischordant, over-compressed, over-processed mush before the listener gets it, it is still vital that the replay equipment reproduce all of this noise, compression and mush accurately.
Because that's what it's meant to sound like.
The artist intended it to sound a particular way. You're free to like it or not, but it'd be better if you judged it on the basis of it being the sound they heard when they said "yes, go with that", rather than a loose approximation of it.
Sadly this is in fact true. Well nearly.
I can think of three occasions on which I can say (I used to design Hifi and other audio gear) that I really understood what true HiFi was.
1/. Testing out radio 3 classical concert on the sort of speakers I could never afford, and hearing not 'applause' but individual people clapping. Clapping is exactly where the time domain is important. Otherwise its sounds like white noise.
2/. Listening to an album called 'Bass Culture' - probably the heaviest and most political reggae album ever made over a 2KW setup featuring 4x15" woofers in closed cabinets. I did manage to crack the studio ceiling. BUT every bass slap was a punch in the guts ..superb.
3/. Standing in the middle of a disco dance floor where we had rigged medium and HF horns to focus the sound exactly there, and actually pull the level down at least 30dB off floor where there was a restaurant. SPL up around 110DB with no distortion and it wasn't painful, just unbelievably clear treble with an amazing stereo image.
Yes, you can do good sound systems, but generally the cost is in 5 figures.
However, post processing does not murder the sound of hi hats and snare drums. Or an acoustic guitar.
Its worth while having the kit if you really can hear the difference. Most people cant.
The best setup I ever heard was an ambisonic array, 8 Kef 107 head units arranged in pairs; front, back, up and down and one very large sub woofer. The recordings were made on a Soundfield microphone in B format on 4 track reel to reel without any post processing. Ambisonics recreates sound pressures measured by a four capsule microphone time aligned to a single point to produce X, Y and Z direction and pressure W. Sadly not a very commercial proposition.
Speakers cant produce them, and the air cant transfer them. The fact that mp3 represents sine waves well and square waves badly is by design. The fact that you can draw a square wave on a computer screen does not mean you will ever hear one.
Just like you wont see a square wave on the ocean
Do you mean 'can't' or do you mean that some information loss occurs? Not quite the same thing, is it? The filtering caused by a well designed speaker or the intervening air is nothing compared to the damage done by a lossy codec.
Posted by someone that doesn't understand Fourier analysis and waveforms.
Square waves (and impulses) are a very useful mechanism in the design of audio systems, it doesn't mean you are trying to design a system in order to reproduce a square wave!
If you understand what square waves are and the harmonics contained within, then you can easily see why they are so useful.
It's true that you can't hear them, but that remark fails to understand what square waves are.
Humans don't hear square waves (or any other waveform), what they hear is the summation of harmonics. A true square wave has an infinite number of harmonics, and we DO hear the harmonics up to around 20Khz. So in a way we do hear the square wave, but we don't hear everything in them, but then it's a physical impossibility for us to hear everything in the square wave.
That may be, but the better a job an audio system can do on square waves, the better it will do on clarinet music, the clarinet being an instrument whose sound is mainly made up of the odd harmonics.
Do the ears hear the summation of harmonics? That would imply localised information processing in the ear. Isn't it that the ear registers pressure and phase differences, and the brain that interprets these as harmonics?
The ear does do local processing. It does not send the raw waveform to the brain. Instead you have sensor cells arranged in the cochlea that get excited by different frequencies, and the brain gets the result of this frequency analysis.
When I crank the gain on my guitar amp and plugin the Les Paul beast, I assure you, you can damn near hear a square wave ;-) (is there a smiley for humorous indignation?)
caveat: Transistor amp of course, valves are a totally different thing, I much prefer valves distorting rather than transistors clipping.
get yourself a function generator, and one of those crappy old speakers, you can quite clearly hear a square wave(well the fundamental & harmonics)... In fact get a bunch of function generators and an oscilloscope and you can make your own square wave! from scratch :-) and then compare it to the one on the FG.
you ears may vary
There is a fair bit of localised processing in the ear, both mechanically and hydraulically in the fluid-canals, and then 'traditionally' within the neural nets that further pre-process the signals from the sensory hairs before going to the brain.
There's a heck of a lot of physically-distributed processing in an animal - for an obvious extreme example, the patellar reflex does not involve the brain at all.
Yep, square waves are fuck-all use for helping to design equipment meant to reproduce music. We already use DSP to correct phase response, flatten frequency response and disrupt standing waves. Not much point trying to improve stereo image at this point, the industry gave up on two speaker solutions years ago.
To reproduce a square wave you need infinite frequency response and instantaneous rise time. Not gonna happen, not useful anyway.
Me too. Gibson R9 & MESA Mark V with MESA 2x 12 vert slant. Square waves a mundo!
You might feed a square wave into a speaker, but it's not going to be very square when it comes out as sound. Even if you could move (and stop) the speaker cone fast enough the air it moves compresses.
@MacroRodent - is that analysis? It's analogous to having an array of filtered microphones feeding into a DSP.
This is exactly the kind of psudo-science crap the article was about.
Guilty of the fallicy of: false equilancy.
But Clarinets and all other reed instruments are predominantly even harmonics.
Brass and strings are odd..
If you put a square wave through a suitably designed high order 20Khz LP Bessel filter it still looks square, just with sloping sides and a bit of corner rounding.
The point is that what is at stake here is how much we hear in the time domain and how much in the frequency.
My experience suggests both. 'Clicky' sounds abound in the environment, and the information content and especially the location is not just assessed by how the relative volume is affected, between ears but by time delay. Mess with that and it doesn't sound 'real' any more.
'Classical' music is relatively free of sharp transients: Rock music is full of them. If all you listen to is opera, go for low distortion labyrinth.
but stick to IB for Led Zeppelin.. ;-)
"It's analogous to having an array of filtered microphones feeding into a DSP."
I don't think so. Each of those microphones would be sending a filtered version of the sound, whereas the nerve cells fire more frequently when their frequencies are heard, and these firings do not resemble any version of the sound waves. (Disclaimer: I'm not an ear expert, but a tinnitus problem made me read at one time more closely about how the ear works. Tinnitus is (or at least some forms of it are) caused by some of these sensor cells getting activated for no reason. Like a stuck pixel in an LCD. That is why one hears a whining sound at a certain fixed frequency, or frequencies).
Does that mean that a clarinet will sound better on clipping transistor amplifiers than on clipping valve ones?
Mine's the one with the hexfets in the left pocket
So what do you use as a baseline during design if not a square wave? That is, if you create a design, test it, and then set about improving that design, and then re test it? How do you know if you have improved your design? You need a reference set of signals to apply to your design, measure the output, modify, re test with the same set of reference input signals. You have to use some sort of standard input signal which you can re-apply in order to determine if you have actually improved your design. That's where square waves come in (and frequency swept sine waves).
Your statement that square waves are F** all use is incorrect.