Hi res music retailers - who do you use?

[malor]

Yup, for 95% of any kind of use, a 60-100 watt receiver would probably totally cover pretty much anybody.

With a powered sub, anyway. And I've owned a seriously lousy receiver in the (distant) past, so they're not guaranteed to be good.

My focus is on results, as long as the resulting stack isn't ugly. Having lots of boxes and looking complex is an artifact, I think, of the 1970s, when that was the only option. Great sound was finally achievable for the first time, but it took lots of expensive gear. In the 2020s, speakers are still expensive, but the electronics usually don't need to be. They can be quite small and simple-looking, while thoroughly outperforming most separates.

Room correction software is a really big deal.

 

[yd]

Oh speakers now are unreal. Shocking what companies are charging and I have a reasonable tolerance for shock.

Interestingly, one of the Weiss products does have some kind of room correction software but I gather it is a bit clunky.

 

[BadtzMaru]

It was a nice ecosystem, but when my SB3 failed, my only realistic replacement option was the Java client, and that sounded awful. I'm not sure if it was Java itself, the computer, or the DAC on that motherboard. I don't remember what I ended up using.

I really liked that kit. I wish Slim Devices hadn't sold themselves.

IIRC, the server software was written in Perl, which was kind of weird. I'm sure it made more sense in 2001. I never got the idea that it wasn't fully open source. Maybe there were proprietary plugins or something?

The LMS community is alive and well (being in the process of being renamed though), see https://forums.slimdevices.com/. I use it every day! (In my case using the docker image under TrueNAS...)

 

[figatry]

There's really no point to this at all. 44KHz/16-bit is perfect. The reason to use 24-bit is for mixing, to keep quantization errors down in the least significant bits, so that the accumulated noise drops out of the signal when it's remastered to 16 bits. And the reason to sample faster than 44KHz is for frequency response above 20KHz. Which you don't want.

As someone who could, in his youth, hear well above 20KHz (I maxed out the Exploratorium tester in San Francisco, so I could hear to at least 26KHz for a few years as a teenager): you don't care about anything up there. It's all whiny, nasty noise. There is nothing musical, it's all awful. Engineers can't hear it and can't mix it. Imagine mosquito whines forever, but much higher pitched. That's the kind of crap that's up there. You don't care about it or want it for any reason whatsoever.

Seriously, just buy 16-bit/44KHz. You will get literally nothing from higher resolution or higher frequency. The CD format is perfect. They got it in one, and unless human hearing changes, there will never need to be a higher standard for digitization of music.

That said, FLAC format supports more than two channels. Any individual channel is fine at 16-bit/44KHz, but you can potentially do surround sound that way. That's nice for the old quadraphonic stuff (Pink Floyd in quadraphonic is quite cool), or if anyone starts doing multichannel stuff now. AFAIC, 16-bit, 44KHz FLAC is the ideal music format. It's bitperfect, runs on pretty much anything, and supports multichannel.

Do not pay extra for anything above 16bit/44KHz. You might as well be lighting money on fire.

44KHz WRT sample-rate has nothing to do with what frequencies you can hear, it's the "resolution" of the audio. That is a fact, not an opinion.

 

[poochyena]

The reason to charge more for FLAC is probably that the files are bigger. The actual meaningful cost to the provider to send you a FLAC vs. MP3 is minuscule, so most of it is price-gouging. FLACs are technically superior, so they charge more.

ah. Yea, I know the difference between the formats, but I know sometimes music will be remastered (is that the right word?) to improve the sound quality, sorta like how film will be re-scanned to a higher resolution. Guessing none of these companies do that much extra work.

 

[etr]

44KHz WRT sample-rate has nothing to do with what frequencies you can hear, it's the "resolution" of the audio. That is a fact, not an opinion.

The sample rate of digital audio puts a definite cap on the frequencies that can be faithfully reproduced.

At the bare minimum, you would want a sample rate at least double your highest target frequency. That would mean a 44 kHz sample rate would be suitable for, at best, 22 kHz frequencies.

Spoiler

To think about why, think about a 44 kHz sound wave. If you sample that sound wave at a consistent 44 kHz, every measurement you take of the sound wave is going to be at the same amplitude. If all your samples are at the same amplitude, then when you send that series of samples to a speaker for output, you are going to get a flat line that will result in silence.

In contrast, if you sample a 22 kHz sound wave at 44 kHz, you will measure equal but opposite amplitudes (unless you have unlikely timing). The ideal would be to have these captured at the peaks and troughs of the sine wave, but even if you are off target, you will at least get differing amplitudes that will produce some sound, even if the volume may be off. (Unless you get unlucky and hit at or near the zeroes, in which case you could wind up with silence or near silence.)

In practice, I understand that it's better to sample "a little more" than twice your target high frequency, and my understanding is that this is where 44 kHz came from (20 kHz x 2 x 1.1).

Spoiler

Sampling a little more means you cannot miss the high frequency audio wave entirely due to bad timing. It does mean that the implied volume of that wave--based on a simple reconstruction from the samples--will rise and fall over time. That's probably mathematically ideal, but is probably considered acceptable because the intent is not so much to capture the high frequency directly, but instead to capture its interplay with lower frequency sounds. From that perspective, mathematically perfect is probably not required.

From that perspective, malor is 100% correct. The sample rate of CD's is sufficient to capture--even if imperfectly--frequencies of up to about 20 kHz, which was chosen since people do not hear very much at or above that frequency (at least directly).

I suspect you could create artificial tests (especially with two simple, specifically chosen frequencies) where someone with good hearing could hear the difference between 44 kHz and some higher frequencies (say 88 kHz or 96 kHz). That said, I suspect even those folks would struggle to pick out the difference listening to a typical composition.

Edit: Corrected mistakes with spoiler tags

 

[malor]

44KHz WRT sample-rate has nothing to do with what frequencies you can hear, it's the "resolution" of the audio. That is a fact, not an opinion.

You completely misunderstood my comment.

A 44.1KHz signal has an absolute upper limit of 22.05KHz. It cannot correctly represent any signal above that frequency. This is fact, not an opinion, it's called the Nyquist Theorem. All properly-made 44.1KHz playback devices have a steep cutoff starting at 20KHz; there's a specific technical term that's escaping me. This prevents aliasing at the highest frequencies from being audible.

Your ability to hear is separate from that. Human hearing is generally limited to 20KHz at most, and almost always much less as we age. As a young person, I could hear above 20KHz, unlike most humans. And, as someone with actual experience with those frequency ranges (to at least 26KHz), I can assure you that there is absolutely nothing up there you want to hear. It's all ringing and whining, all of it unpleasant. Mixing engineers can't work with it because they can't hear it.

There is, in other words, zero reason to increase sampling rates past 44.1KHz. This already exceeds the range of human hearing. A 96KHz sample will let you sample up to 48KHz audio waveforms, but this is ridiculous. Nobody can hear it. There is no point. It's snake oil, pretend audiophile garbage for people who want to think they're superior to everyone else.

Everything I'm saying here, btw, is factually accurate. If you think anything I'm saying here is wrong, you misunderstand how digital sampling works.

 

[malor]

From that perspective, malor is 100% correct. The sample rate of CD's is sufficient to capture--even if imperfectly--frequencies of up to about 20 kHz,

CDs are literally perfect from 20KHz and down. If a CD cannot reproduce an audio waveform exactly, then that waveform has frequency components above 20KHz.

I'm blurry on the details, I don't actually know the math, but basically a CD player is constantly solving a waveform equation, drawing curves such that the lines fall perfectly on the sample points. Out near the limit of the Nyquist Theorem, I believe that you can end up with multiple curves that fit the available sampling points; you get high frequency aliasing, as the player sometimes makes the wrong choice out of several. That's why there's that rolloff at 20KHz, because below that point, the only valid possible waveform solution that fits the provided samples is the original. Any errors will be in frequency components above 20KHz, and will be suppressed from the output.

 

[malor]

ah. Yea, I know the difference between the formats, but I know sometimes music will be remastered (is that the right word?) to improve the sound quality, sorta like how film will be re-scanned to a higher resolution. Guessing none of these companies do that much extra work.

That's actually sometimes true. The advent of digital music has allowed all kind of bullshit to get pulled with mixing, frequently pushing audio above the clipping point to increase apparent loudness on the radio. (a lookup term for more info: "The Loudness Wars".) You can't pull that kind of crap with vinyl, because if you try, the needle will jump the groove.

If the 96KHz FLAC version is actually the vinyl mix, it can sound a lot better than a mixed-past-clipping 44.1KHz. That's part of why people still like vinyl. A crappy reproduction of a superior mix can sound better than a perfect reproduction of a shit mix.

But that's not the 96KHz or the 24 bits, that's just correct mixing. Do a quality resample of that superior mix down to 44.1/16, and it will take a lot less space, and still sound perfect.

 

[SuperDave]

The only reason for higher sampling rates than 44 is to ease the strain on low-pass filters in mastering; shit happens when you have to cram the whole filter into ~2KHz of sampling signal. Either way, it all boils down to the quality of the master (first, and for the love of God most important), the equipment quality between the source and the transducer (damn near an Easy Button these days, that problem is solved even in Class D and freaking dongle DACs plugged into your phone for mobile use), the quality of the transducer (speakers still have "voice;" your preference is not mine) and (took me a while to learn this, because I'm not young) the room correction. And (this is what I had the hardest time accommodating) software room correction is mature tech. You haven't heard what your equipment can do until you've added that layer. This, too, is becoming an Easy Button, Your existing equipment is very likely a lot better than you think.

But, again, most importantly as others have either stated or alluded to, it all hinges on the quality of the original master. These days that quality - as a whole - is better than it was 50 years ago when I started buying music, but capitalists gonna capitalize and there are (and will be) fer-shit masters of stuff you want to listen to because it's cheaper to produce and they don't care because people participating in this conversation are outliers.

 

[yd]

So you have decent gear, but you don't have room correction in anyway - only relatively intelligent positioning/toeing in/rugs yada yada to help.

But you are using an htpc as your audio source. Is there some room correcting 'system'/device you can get that would operate htpc side pre dumping out your signal to your d2a/pre/amp/speakers that would be useable/buyable?

 

[malor]

So you have decent gear, but you don't have room correction in anyway - only relatively intelligent positioning/toeing in/rugs yada yada to help.

But you are using an htpc as your audio source. Is there some room correcting 'system'/device you can get that would operate htpc side pre dumping out your signal to your d2a/pre/amp/speakers that would be useable/buyable?

As far as I know, room correction requires DSP chips. I don't think any of those algorithms can run on a standard CPU. They're not fast enough.

 

[etr]

CDs are literally perfect from 20KHz and down. If a CD cannot reproduce an audio waveform exactly, then that waveform has frequency components above 20KHz.

"Perfect" is too imprecise a term here.

I'll take it two steps further: (1) you cannot reproduce a mathematical match for a 20 kHz wave form of changing amplitude with (about) two samples per wavelength and (2) sensitive ears might be able to hear that difference, given an isolated example.

That said, the equipment is used for playing the very mixed waveforms of music, game audio, and movie sound. Whatever differences a sensitive ear might be able to discern in isolation are going to get lost in that scenario.

All that said, if someone wants to "fix" that isolated scenario to satisfy themselves that the are getting audio at or above the limits of their hearing, well they're free to cause the end of that rainbow.

That said, it would be a little silly to buy that format just to put it in a lossy format like MP3 for regular listening.

Spoiler

I should phrase myself better, as "perfect" could mean "mathematically identical waveform" (I'll call that "m-perfect") and "a sufficiently identical waveform that one can't hear the difference" (I'll call that a-perfect).

CD audio is probably practically a-perfect for all but the most contrived scenarios (and probably the most sensitive ears).

Let's really stress it, though. Let's imagine the following, with variable amplitude.

(1) An artificial "square wave" at 20 kHz
(2) An artificial "triangle wave" at 20 kHz
(3) A sine wave at 20 kHz

I gather that there are frequencies where one can hear the difference between a square wave, a triangle wave, and a sine wave. (The square wave and triangle wave were options used in PCM audio, and their use is considered one of the reasons that retro games that used PCM audio has a distinctive sound.)

That doesn't mean that anyone could tell the difference at 20 kHz, but for the sake of the thought experiment let's assume you could in your teenage years, at least in isolation.

That said, distinguishing between the above cases to come up with an m-perfect wave form would be a non-starter. An output device could probably tell these apart and match the wave form if the amplitude (volume) was constant, but significant amplitude variation is probably going to make even that tough. And, if amplitude variation makes it impossible to distinguish between triangle waves and sine waves, it's also not enough to perfectly nail an m-perfect sine wave, either.

Which is fine. The m-perfect standard is going to provide zero benefit over a-perfect (by definition of the later), so any money spent attaining it is likely wasted. And, even something that is not a-perfect in the isolated case probably us in the real use case

I'm blurry on the details, I don't actually know the math, but basically a CD player is constantly solving a waveform equation, drawing curves such that the lines fall perfectly on the sample points. Out near the limit of the Nyquist Theorem, I believe that you can end up with multiple curves that fit the available sampling points; you get high frequency aliasing, as the player sometimes makes the wrong choice out of several.

I believe the term is "interpolation". The results are probably not mathematically perfect in most cases, but probably much better than folks are likely to hear.

And, unless someone mistakes me for a proponent of audiophilesgold conductor Ethernet cables, I listen to MP3's. But just like I'm going to argue that that the A-B testing rejected by audiophiles is a good thing, I will also insist on pointing out that an interpolated wave form form CD audio is (mathematically) "perfect". It's not; it's just that the differences are small enough that folks aren't generally going to notice.

And, FWIW, higher sampling rates are nowhere near the level of snake oil of gold conductor Ethernet cables.

 

[malor]

I believe the term is "interpolation". The results are probably not mathematically perfect in most cases, but probably much better than folks are likely to hear.

No, it's not interpolation, it's actually solving a wave equation. Each new sampling point generates a new curve, such that the "drawn" output waveform will smoothly curve to pass over that sampling point. As new sampling points are presented, the smooth, generated curve keeps shifting, creating an audio waveform. I don't know the actual algorithm, but I know there is one. It's not a stuttery sample, where each data point is taken as a precise volume level at that precise instant, it's a constantly shifting curve that's always including, IIUC, at least three samples, possibly more.

A properly working CD player will exactly duplicate the original input waveform. If it does not, then the input waveform has components above 20KHz, full stop. It's not just raw samples, there's a math layer involved.

you cannot reproduce a mathematical match for a 20 kHz wave form of changing amplitude with (about) two samples per wavelength

That's why they sample above 40KHz for a 20KHz output. DVDs give a little more headroom, at 48KHz. There's no reason to go any further. The additional gain is literally and exactly zero.

 

[yd]

No, it's not interpolation, it's actually solving a wave equation. Each new sampling point generates a new curve, such that the "drawn" output waveform will smoothly curve to pass over that sampling point. As new sampling points are presented, the smooth, generated curve keeps shifting, creating an audio waveform. I don't know the actual algorithm, but I know there is one. It's not a stuttery sample, where each data point is taken as a precise volume level at that precise instant, it's a constantly shifting curve that's always including, IIUC, at least three samples, possibly more.

A properly working CD player will exactly duplicate the original input waveform. If it does not, then the input waveform has components above 20KHz, full stop.



That's why they sample above 40KHz for a 20KHz output. DVDs give a little more headroom, at 48KHz. There's no reason to go any further.

When I read this I think, Malor most certainly is a proponent of a Class A amplifier

100 watts probably covers it amiright?

 

[malor]

When I read this I think, Malor most certainly is a proponent of a Class A amplifier

100 watts probably covers it amiright?

I have no strong opinions on amplifiers. Well, I have one, that I prefer Denon's tech to Onkyo's, because Onkyos run physically hot and audibly bright. But that's an obsolete opinion, as Onkyo failed awhile back.

But in terms of tech, I don't care. Class D amps sound fine to me. Most receivers sound excellent.

I prefer sealed subs to ported, and the sub is where most of your amplifier power should be.

 

[yd]

I have no strong opinions on amplifiers. Well, I have one, that I prefer Denon's tech to Onkyo's, because Onkyos run physically hot and audibly bright. But that's an obsolete opinion, as Onkyo failed awhile back.

But in terms of tech, I don't care. Class D amps sound fine to me. Most receivers sound excellent.

I prefer sealed subs to ported, and the sub is where most of your amplifier power should be.

I will say, I prefer a class a/b but with a reasonably high bias into class A. Not really interested in class D. I DEFINITELY prefer a sealed speaker to a ported one. Almost a deal breaker for me to be honest.

 

[etr]

A properly working CD player will exactly duplicate the original input waveform. If it does not, then the input waveform has components above 20KHz, full stop. It's not just raw samples, there's a math layer involved.

I am skeptical that the the reproduced wave form is exact from a mathematical perspective, but I will surrender because I lack the math background to prove otherwise. (Heck even if I had the math background, it would probably be difficult to distill it down enough to prove the point, anyway.)

That said, I am reasonably convinced that the value of higher sampling rates for a 20 kHz frequency cap is limited.

 

[SuperDave]

I will say, I prefer a class a/b but with a reasonably high bias into class A. Not really interested in class D. I DEFINITELY prefer a sealed speaker to a ported one. Almost a deal breaker for me to be honest.

We're at a technological place, from both a hardware and software point of view, where good music is cheap and small. All of my amplification is Class D - the desktop amp for my LS50's sits under the monitor and the pair driving the R3's (it's all vanity, one would more than suffice but I wanted BTL again, in my system just because) are 200x350mm, run at room temp while outputting a kilowatt each into the load with SINAD which is barely measurable. And $1500 covered all three. For your tastes (and situation, I get your need to occupy space ), if you upgrade look to Benchmark. Their AHB2 amp simply negates the need to spend more money, albeit at $3.5k they're beyond my budget but you don't have to spend $10k for this kind of stupid quality any more.

Porting is relative and situational depending on room, location and usage (and now, software room correction). My Hsu sub is plugged tight but all the speakers are open ported; that's what works in my spot for my ears. YMMV, and spend the time experimenting. Especially correction.

 

[Schpyder]

Room correction software is a really big deal.

Having recently purchased one of the newer Pioneer Elite AVRs with built-in Dirac Live correction (and thank god I actually had a boom mic stand hanging around to let me measure at all the required locations), I really cannot emphasize this enough. Good room correction is game-changing.

A properly working CD player will exactly duplicate the original input waveform. If it does not, then the input waveform has components above 20KHz, full stop. It's not just raw samples, there's a math layer involved.

FWIW, it's almost certain that the real, actual physical waveform always has ultrasonic components, it's just that they're pointless for reproduction for human hearing. It's ALSO worth noting that the vast majority of studio recording gear isn't really designed for ultrasonic recording either. You're going to need the entire signal chain, from microphone hardware, every stage of hardware processing, and audio interface (this is a big one, most interfaces don't accurately record signals above 30kHz or so; high sampling rates in those are almost strictly for avoiding aliasing in audible/near-audible frequencies) all need to be designed for proper recording of ultrasonics. Almost nothing is actually recorded with that in mind, which is one of the main reasons chasing high sampling rates is so pointless.

 

[malor]

Almost nothing is actually recorded with that in mind, which is one of the main reasons chasing high sampling rates is so pointless.

Yep, the more closely you examine the problem, the more reasons it doesn't really work keep cropping up.

It's an idea so bad that it's fractal.

 

[figatry]

You completely misunderstood my comment.

A 44.1KHz signal has an absolute upper limit of 22.05KHz. It cannot correctly represent any signal above that frequency. This is fact, not an opinion, it's called the Nyquist Theorem. All properly-made 44.1KHz playback devices have a steep cutoff starting at 20KHz; there's a specific technical term that's escaping me. This prevents aliasing at the highest frequencies from being audible.

Your ability to hear is separate from that. Human hearing is generally limited to 20KHz at most, and almost always much less as we age. As a young person, I could hear above 20KHz, unlike most humans. And, as someone with actual experience with those frequency ranges (to at least 26KHz), I can assure you that there is absolutely nothing up there you want to hear. It's all ringing and whining, all of it unpleasant. Mixing engineers can't work with it because they can't hear it.

There is, in other words, zero reason to increase sampling rates past 44.1KHz. This already exceeds the range of human hearing. A 96KHz sample will let you sample up to 48KHz audio waveforms, but this is ridiculous. Nobody can hear it. There is no point. It's snake oil, pretend audiophile garbage for people who want to think they're superior to everyone else.

Everything I'm saying here, btw, is factually accurate. If you think anything I'm saying here is wrong, you misunderstand how digital sampling works.

I didn't misunderstand. We just disagree. 44.1KHz sample rate means that every second 44,100 samples of audio are being recorded or played, 96KHz sample rate means that 96,000 samples of audio are being recorded or played. Which one is going to be a truer representation? The one with 41,000 or 96,000 samples per second?

 

[kaveesak]

Yup, for 95% of any kind of use, a 60-100 watt receiver would probably totally cover pretty much anybody.

I'm in the 95% and who are the 5%?)

 

[ant1pathy]

I didn't misunderstand. We just disagree. 44.1KHz sample rate means that every second 44,100 samples of audio are being recorded or played, 96KHz sample rate means that 96,000 samples of audio are being recorded or played. Which one is going to be a truer representation? The one with 41,000 or 96,000 samples per second?

100% identical through 20.5k hertz !

 

[Schpyder]

I'm not sure why this is such a weird sticking point for audio. You don't see people complaining that their TV isn't putting out enough infrared and ultraviolet spectrum and that they're somehow missing out on wavelengths that they absolutely can't see.

 

[figatry]

100% identical through 20.5k hertz

Ok, I would like to ask the question again, but let's say you record two indentical sine wave tone's(pick your frequency), but they are not in sync with each other, what is your answer then?

 

[whoisit]

Let's consider a spherical cow on a flat plane of space time. Does adding a second spherical cow change the number of angels dancing on the head of a pin.

 

[ant1pathy]

Ok, I would like to ask the question again, but let's say you record two indentical sine wave tone's(pick your frequency), but they are not in sync with each other, what is your answer then?

I'm sorry, I don't understand the question...?

 

[figatry]

I'm sorry, I don't understand the question...?

There are two sources (A and B) of sound waves at the same frequency, one started later than the other and they are not oscilating in sync with each other and are 90 deg out of sync. We will assume that source A is in sync with the analog to digital converter which is sampling at the frequency of the sine wave and and it is a "perfect" reproduction. Will source B be sampled?

 

[etr]

There are two sources (A and B) of sound waves at the same frequency, one started later than the other and they are not oscilating in sync with each other and are 90 deg out of sync. We will assume that source A is in sync with the analog to digital converter which is sampling at the frequency of the sine wave and and it is a "perfect" reproduction. Will source B be sampled?

For a constant amplitude waves at 20 kHz (you should be filtering out frequencies above that for CD audio), both waves would be sampled. CD audio samples at more than twice the highest supported frequency (20 kHz) to ensure you cannot get entirely bit by unfortunate sample timing. Over the course of 5 cycles, you will have gotten samples at multiple points on the waveform.

 

[etr]

I'm not sure why this is such a weird sticking point for audio. You don't see people complaining that their TV isn't putting out enough infrared and ultraviolet spectrum and that they're somehow missing out on wavelengths that they absolutely can't see.

TV's already don't reproduce much of the visible spectrum. They produce specific wavelength ranges in mixes that tickle our cones and rods in a manner similar to a full visual spectrum image without reproducing the full visual spectrum.

In short, the point can be more or less explained with an overview of implementation and the human visual system without getting into the math.

Spoiler

There are probably mixed positions on the audio sampling question. Personally, I was not questioning whether there was much point in trying to capture frequencies above 20 kHz. I was questioning whether more samples could render more mathematically accurate reproduction of 20 kHz and below. (A related question is whether anyone could hear the difference, which I'm skeptical of for typical listening scenarios ).

The answer I'm hearing from some folks in thread is, "No, CD audio gives perfect reproduction from 20 kHz on down, therefore there is no reason for higher sample rates." I'm not sold on the contention, but I cannot refuse or convince myself of this point without getting into the math, which is on me if I want to go there.

 

[figatry]

For a constant amplitude waves at 20 kHz (you should be filtering out frequencies above that for CD audio), both waves would be sampled. CD audio samples at more than twice the highest supported frequency (20 kHz) to ensure you cannot get entirely bit by unfortunate sample timing. Over the course of 5 cycles, you will have gotten samples at multiple points on the waveform.

I was asking someone else. Additionally, you didn't answer the question as asked and introduced trivial information. This isn't rocket science.

 

[Baenwort]

I'm in the 95% and who are the 5%?)

My son and his fixation on music that has as its only spoken word "bass" and the house shaking sound that takes at least 1200W to produce from my setup. Most of the wattage goes to the transducers (Buttkickers) on the couch and the 12" sealed sub.

I'm pretty sure all the frequencies above 130HZ only use 1% of that as I have a high efficiency horn setup and tower midwoofers. (~98db/W)

Once a week he gets to make the family wear ear protection and crank it up. Otherwise my wife and I are the sound police with my dB meter and limit him to 85dB (C weight, slow averaging)

 

[yd]

I'm in the 95% and who are the 5%?)

A good house party or 'lets see what this can do' session can be occassions where you suddenly find your 60 watt receiver solely laking - or in my case, that NAD amplifier going 'pop' and so much for that party (which was also with fairly easily driveable speakers). Soft clipping my ass - and that was an actual power amplifier, not a receiver.

 

[Pericles]

There are two sources (A and B) of sound waves at the same frequency, one started later than the other and they are not oscilating in sync with each other and are 90 deg out of sync. We will assume that source A is in sync with the analog to digital converter which is sampling at the frequency of the sine wave and and it is a "perfect" reproduction. Will source B be sampled?

That seems like a bit of a misunderstanding about what is the result of adding two signals. You don't get to record both sine waves as independant entities but the sum of them.
In a context where you're producing a signal to put on a CD, whichever part of the input signal lies below 20kHz will be faithfully encoded, the rest won't be recorded at all*.

* my 20kHz low-pass filter is perfect

 

[yd]

We're at a technological place, from both a hardware and software point of view, where good music is cheap and small. All of my amplification is Class D - the desktop amp for my LS50's sits under the monitor and the pair driving the R3's (it's all vanity, one would more than suffice but I wanted BTL again, in my system just because) are 200x350mm, run at room temp while outputting a kilowatt each into the load with SINAD which is barely measurable. And $1500 covered all three. For your tastes (and situation, I get your need to occupy space ), if you upgrade look to Benchmark. Their AHB2 amp simply negates the need to spend more money, albeit at $3.5k they're beyond my budget but you don't have to spend $10k for this kind of stupid quality any more.

Porting is relative and situational depending on room, location and usage (and now, software room correction). My Hsu sub is plugged tight but all the speakers are open ported; that's what works in my spot for my ears. YMMV, and spend the time experimenting. Especially correction.

Now I am not going to yuck a yum, but the claim to fame of the AHB2 is essentially 0 THD (well, 0.0003) - not sure what it does at higher output or how that looks across full frequency range.

That said, THD even if totally eliminated is not the full story; a timely release from PS Audio today just for instance


View: https://www.youtube.com/watch?v=EtUC7mLOEF0


He is talking about distortion waaay higher than my amps at 0.08%; my antique amps are basically 0.002 rising at higher frequencies to say 0.005. But as he says, that doesn't tell us how the amp actually sounds. My 100 watt Schitt amp goes to pieces at higher output levels on the low end (edit - and it has similar but not quite as good thd but certainly not enough grunt imho), my 250 watt Jeff Rowland is bomb proof at even higher levels. Different designs are going to sound different, even with identical THD. For myself as mentioned, I want something heavily biased in class A switching to A/B when its party time for maintaining control/low end slam.

There is no way to hear this AHB2 here locally I don't think sadly - that said, it wouldn't pass WAF test and I'd be concerned that they have temp indicators on the front aka methinks you could overheat it or more relevantly, I think I could overheat it in my listening environment.

 

[figatry]

That seems like a bit of a misunderstanding about what is the result of adding two signals. You don't get to record both sine waves as independant entities but the sum of them.
In a context where you're producing a signal to put on a CD, whichever part of the input signal lies below 20kHz will be faithfully encoded, the rest won't be recorded at all*.

* my 20kHz low-pass filter is perfect

That didn't answer the question.

That seems like a bit of a misunderstanding about what is the result of adding two signals. You don't get to record both sine waves as independant entities but the sum of them.
In a context where you're producing a signal to put on a CD, whichever part of the input signal lies below 20kHz will be faithfully encoded, the rest won't be recorded at all*.

* my 20kHz low-pass filter is perfect

That didn't answer the question.

 

[Wheels Of Confusion]

That didn't answer the question.

The question doesn't really make sense as posed.

 

[Schpyder]

There are two sources (A and B) of sound waves at the same frequency, one started later than the other and they are not oscilating in sync with each other and are 90 deg out of sync. We will assume that source A is in sync with the analog to digital converter which is sampling at the frequency of the sine wave and and it is a "perfect" reproduction. Will source B be sampled?

Tell me you don't understand frequency analysis, time domain waveform analysis, or digital signal processing without telling me.

To make it simple for you: superposition of two sine waves at the same frequency is a sine wave at that same frequency, regardless of phase shift or amplitude difference between the two original waveforms. This is easily mathematically provable via the compound angle formula. But this betrays your misunderstanding. If you're sampling at the same frequency as the waveforms, neither will be accurately reproduced. You're getting one sample per cycle, at the same point on the waveform every time. You are, in effect, recording a flat line.

Now, I assume you meant to say that you are sampling at twice the frequency of source A, so that you would be perfectly sampling the peaks and valleys of said signal. Also, I assume you are aware that modern statements of the Nyquist limit state it as a strict inequality: that is that you must be sampling at greater than 2x the maximum frequency you wish to reproduce, to avoid aliasing errors at that frequency. What am I saying, of course you don't, because that's exactly the situation you've posed here. It's also why we low-pass filter at a frequency a good bit below the Nyquist cutoff: to avoid the reconstruction errors that occur as you get close to the critical frequency.

 

[etr]

I think I've answered most of my concerns on CD audio, even without getting into the math. I'll concur that the 44.1 kHz sample rate should do an audibly perfect job for 20 kHz and below sine waves. There is likely no practical point to higher sample rates for 20 kHz and below audio.

Spoiler

As noted before, 44.1 kHz will adequately sample a constant volume 20 kHz sine wave in about 5 cycles.

The thing I was hung up on was that for a set of samples from such a constant volume wave, I suspect I could produce another 20 kHz wave that would match by varying amplitudes. If I can have two (or more) wave for represented by a single set of samples, then a reconstruction algorithm isn't really going to know which one is right. That would take mahematical perfection off the table.

However, I thought about the timescale involved. 5 cycles of a 20 kHz wave amounts to a quarter of a millisecond. I think it's a safe bet that no one can audibly detect a change in amplitude (volume) on anything approaching a quarter of a millisecond. It's a safe bet that what we perceived as volume is a running average of amplitude over time.

TLDR: I'd pay extra for lossless formats I could generate into my preferred compressed formats for regular listening, I wouldn't pay for more samples (or larger samples, since I would be listening and not remixing).