TR Forums

just brew it! wrote:

Nec_V20 wrote:

I wouldn't say I was tone deaf, but I don't have an absolute ear - or anywhere close to it - so I am perfectly happy with the on-board sound on my Gigabyte X58A-UD3R.

The problem with badly implemented onboard typically isn't a pitch issue; it is noise and/or distortion. But (as with artifacts from lossy compression algorithms like MP3), some people are more bothered by it than others.

I'm not sure what kind of noise you're referring to, but if it's noise routed to the speakers through the audio parts at the back of the board, I don't think there's a problem unless you're listening at really high volume through some huge speakers. What bothers me is when your front panel audio ports are unshielded and they're located near the front panel USB ports. This is applicable to both onboard audio as well as discrete audio cards as long as the front panel audio connectors are unshielded. Every time I access, say, a flash drive on those USB ports I can hear those buzzing noises which are clearly interference from the USB ports making it to the audio ports.

As for distortion, is THD responsible for the 'liveliness' or lack thereof of sound quality? Or the crispness of every instrument when listening to music? Are SNR and THD+N the only things we're looking for? How about frequency response? Note that I didn't even include date rates; HD Audio can support up to 32-bit/192KHz (most implement 24-bit/192KHz though), which is awesome on paper but onboard audio obviously can't touch dedicated cards with far more legroom for quality analog components. So obviously, merely having high 24-bit/192KHz sampling resolution doesn't tell you everything about sound quality. In fact, the Realtek control panel has some settings to switch between different sampling rates, and I honestly can't tell any difference between 16-bit/44.1KHz and 24-bit/192KHz.

ronch wrote:

In fact, the Realtek control panel has some settings to switch between different sampling rates, and I honestly can't tell any difference between 16-bit/44.1KHz and 24-bit/192KHz.

If the source file is 16/44.1 it'll only be resolved as 16/44.1, at least by onboard sound chips. Resampling is possible and boxen do exist to do it, but it's overkill for onboard. Try to find a 24/192 source file and compare each at the native resolution.

ronch wrote:

I'm not sure what kind of noise you're referring to, but if it's noise routed to the speakers through the audio parts at the back of the board, I don't think there's a problem unless you're listening at really high volume through some huge speakers. What bothers me is when your front panel audio ports are unshielded and they're located near the front panel USB ports. This is applicable to both onboard audio as well as discrete audio cards as long as the front panel audio connectors are unshielded. Every time I access, say, a flash drive on those USB ports I can hear those buzzing noises which are clearly interference from the USB ports making it to the audio ports.

Hence my recommendation (several posts back) to replace the front panel wiring or switch to rear jacks if noise is a problem.

ronch wrote:

As for distortion, is THD responsible for the 'liveliness' or lack thereof of sound quality? Or the crispness of every instrument when listening to music?

It could certainly be perceived as such. At low levels, the effect is subtle and may not even be perceived as distortion per se. At higher levels it tends to be perceived as a "dirtiness" or fuzziness, or may give a harsh edge to the sound.

Think of it this way: The tone of an overdriven electric guitar (as in hard rock / heavy metal) is what you get when you take a clean guitar tone and intentionally introduce large amounts of harmonic distortion.

ronch wrote:

Are SNR and THD+N the only things we're looking for? How about frequency response?

Frequency response is, of course, very important. You want to reproduce everything that was in the original source material. It is not uncommon (even today) for there to be some low frequency rolloff when driving low impedance loads (like headphones), due to under-speced DC blocking capacitors in the output circuit. Fortunately some judicious application of EQ can generally make up for this.

ronch wrote:

Note that I didn't even include date rates; HD Audio can support up to 32-bit/192KHz (most implement 24-bit/192KHz though), which is awesome on paper but onboard audio obviously can't touch dedicated cards with far more legroom for quality analog components. So obviously, merely having high 24-bit/192KHz sampling resolution doesn't tell you everything about sound quality. In fact, the Realtek control panel has some settings to switch between different sampling rates, and I honestly can't tell any difference between 16-bit/44.1KHz and 24-bit/192KHz.

As the Cap'n has noted, most of your source material is probably 16/44 to begin with, so bumping the rate up doesn't buy you anything meaningful.

just brew it! wrote:

ronch wrote:

As for distortion, is THD responsible for the 'liveliness' or lack thereof of sound quality? Or the crispness of every instrument when listening to music?

It could certainly be perceived as such. At low levels, the effect is subtle and may not even be perceived as distortion per se. At higher levels it tends to be perceived as a "dirtiness" or fuzziness, or may give a harsh edge to the sound.

Think of it this way: The tone of an overdriven electric guitar (as in hard rock / heavy metal) is what you get when you take a clean guitar tone and intentionally introduce large amounts of harmonic distortion.

And then we get into the "order" of the distortion, even or odd. Tubes distort in even-order, which sounds good to the ear (and is why really good guitar amps still use tubes) while sold-state distorts in odd-order, which sounds discordant to the ear. This property of distortion explains the oddest onboard audio setup ever, the AOpen AX4b 533, powered by a Sovtek 6922 triode vacuum tube. As if a Socket 478 P4 heating up the case wasn't enough, now there was a firebottle involved (tubes run very very hot; touch one that's operating and you'll wish you didn't).

LOL, a vacuum tube on an ATX motherboard?

You dawg, I heard you liked analogue so we're gonna make a steam-powered motherboard next.

I wonder how the new Realtek ALC 1150 audio codecs, with 115dB SNR DAC, Twin TI NE5532 Amps, one as a differential amp and the other as a premium headset amp! Add to that cap-less Direct Drive tech, EMI shielding, PCB isolate shielding and DTS connect is going to perform compared to the most popular sound cards Asus Zonar etc. Its very hard to find specs on the new chip. I mean realteks site does not even list it, it stops at ALC889!

The specs look great...I guess the implementation on each manufacturers board is going to make a difference.
I have been using HDMI audio out through my vid cards and G930 wireless USB headset with simulated surround that has worked flawlessly on all 3 machines in my house. They cannot take too much bass at high volumes, but i can say they have some bass to them. Its just that the 40mm drivers can get overloaded. I also love that they block out 25db maybe more of sound just putting them on and they are very comfortable on my pea sized head.
I have heard so much bad things about USB audio solutions..I guess i just got lucky or just do not use the programs they mess up on.

vargis14 wrote:

I wonder how the new Realtek ALC 1150 audio codecs, with 115dB SNR DAC, Twin TI NE5532 Amps, one as a differential amp and the other as a premium headset amp! Add to that cap-less Direct Drive tech, EMI shielding, PCB isolate shielding and DTS connect is going to perform compared to the most popular sound cards Asus Zonar etc. Its very hard to find specs on the new chip. I mean realteks site does not even list it, it stops at ALC889!

The specs look great...I guess the implementation on each manufacturers board is going to make a difference.

Interesting, I hadn't heard about that. I did find this press release on ASRock's site though. A few thoughts:

• As you've already noted, it really is going to depend on the individual motherboard manufacturer's implementation. I find it odd that they don't quote THD specs for something that is being pitched as an audiophile solution; they also appear to be quoting the theoretical S/N of the DAC alone, and don't actually say what the S/N of their implementation is.
• There's nothing magical about making an audiophile grade PC audio codec; VIA's Envy chips (used in M-Audio's pro audio products) are great, and they've been around for at least a decade.
• There's nothing magical about the TI NE5532 either. It's an average op amp chip that has been around for more than 3 decades, and costs about 30 cents apiece in quantity. I also seriously doubt it has anything approaching 115dB S/N ratio, especially when used in an environment like a PC motherboard.
• Don't get me wrong, the balanced capless output stage is a good thing. However, to imply that this is responsible for improving the bass response is a bit disingenuous. Spending another 10 cents on better caps (without needing to resort to the capless design) would also improve the bass response. But "we decided not to cheap out on the DC blocking caps" doesn't make for particularly good marketing copy!

vargis14 wrote:

I wonder how the new Realtek ALC 1150 audio codecs, with 115dB SNR DAC, Twin TI NE5532 Amps, one as a differential amp and the other as a premium headset amp! Add to that cap-less Direct Drive tech, EMI shielding, PCB isolate shielding and DTS connect is going to perform compared to the most popular sound cards Asus Zonar etc. Its very hard to find specs on the new chip. I mean realteks site does not even list it, it stops at ALC889!

The Realtek site stops at ALC892. I guess it's kinda sad that most decent motherboards today implement the ALC892. Not that it's a poor codec; it's just that the ALC889 actually has better specs if you take some time and compare the two. Of course it all depends on how the board maker uses the codec and 108dB probably isn't achieved often with most ALC889 implementations, but still, a fall from 108dB to 95dB is a sign that the ALC892 is Realtek's answer to the board makers' requests to make a cheaper codec. Besides, it's not like most folks care about audio quality these days anyway. As long as they can hear sound that isn't totally garbled, they're ok with it.

While integrated audio is a big advantage for most folks in that any PC you buy today automatically comes with 'free' quality HD audio, it's also unfortunate that it has made people not look beyond onboard sound.

I've been talking to some friends about PC audio lately, actually, and it seems all the folks I've talked to are quite contented with onboard audio, with all but one guy having no plans at all to step up to even a cheap but proper audio card. One even went so far as to say that a dedicated sound card is not necessary because, where he lives, big, nice speakers are the norm. How mistaken can that be?

My living room PC outputs digital audio through the HDMI cable. I wouldn't expect a sound card to equal the DAC quality of my Onkyo receiver.

However, my gaming PC has analog speakers and headphones plugged into it. I've been putting Sound Blaster cards into my gaming PCs for a long time. The latest one is a Sound Blaster ZX.

just brew it! wrote:

There's nothing magical about making an audiophile grade PC audio codec; VIA's Envy chips (used in M-Audio's pro audio products) are great, and they've been around for at least a decade.

Correct me if I'm wrong, but according to VIA's product page, their Envy product line is composed of PCI audio controllers, not audio codecs. These controllers don't seem to specify SNR ratings as well because these are strictly audio controllers that do what the CPU does in a purely onboard audio setup. They need a separate audio codec, probably from VIA as well. All the low end VIA-based sound cards I've seen such as the Vantec UGT-S100 come with a VIA audio controller (the Tremor, in this case) and a VIA audio codec. The codecs specify the SNR since it's where the DACs reside, as I understand it.

Btw, most VIA Envy-based audio cards are used in expensive implementations, apparently. At just $21 though, this Vantec card right here seems like an impulse buy. I wonder just how good this thing sounds. Of course, I don't think anybody will buy this expecting it to blow their onboard audio away, but still, I'm curious. Not that I would actually buy one, of course, with my X-Fi just a whim away.

ronch wrote:

The Realtek site stops at ALC892. I guess it's kinda sad that most decent motherboards today implement the ALC892, not that it's a poor codec; it's just that the ALC889 actually has better specs if you take some time and compare the two.

It looks like a lot of Asus' boards use either the 892 or the 887 these days. All three chips are similar, except that (as you point out) the 889 has better S/N, distortion, etc. specs. I haven't done a detailed comparison of all other features to see exactly *how* similar these three chips are; it may be a simple binning thing, where chips whose DAC and/or ADC circuits don't quite make the cut as an 889 get re-branded as an 887 or 892.

ronch wrote:

Correct me if I'm wrong, but according to VIA's product page, their Envy product line is composed of PCI audio controllers, not audio codecs.

Yeah, my bad, I must have been thinking of their Vinyl HD codec line, not Envy. But their codecs do have better specs than Realtek's (the 1150 notwithstanding), so the 1150 is in a sense just Realtek getting caught up to (and finally surpassing if the specs are to be believed) where VIA was at years ago.

It's pretty rare to see a VIA codec on a motherboard anymore. My old Asus M3A78-CM and M4A78T-E boards both had VIA codecs on them, but newer boards seem to be all Realtek.

Edit:

ronch wrote:

Btw, most VIA Envy-based audio cards are used in expensive implementations, apparently. At just $21 though, this Vantec card right here seems like an impulse buy. I wonder just how good this thing sounds. Of course, I don't think anybody will buy this expecting it to blow their onboard audio away, but still, I'm curious. Not that I would actually buy one, of course, with my X-Fi just a whim away.

The Chaintech AV-710 was another inexpensive (and quite popular) Envy based discrete audio card back in the day. The card is no longer produced, and Chaintech has been totally absent from the consumer PC market for a number of years. I guess the company is still around in some form and (as of about a year ago) was planning to return to the motherboard and graphics card business.

Wow, it's been ages since I've posted anything. But since I was thinking about this topic anyways, I figured why not.

I've been pumping out DTS connect from my x79 pro to an old receiver that I put in the office. And honestly, I couldn't be happier. I know that doesn't really answer your question, especially if you're thinking analog, but after years of thinking I wanted to go back to a good dedicated soundcard (and try to skip the driver hassles if possible), I've found that DTS Connect /DD Live type options work pretty well for me.

I've got my 5.1 in games, and I've got my clean audio to good components.

So, do I miss 3rd party soundcards?


Yes.

I miss my AWE64 Gold I miss having 32 to 64 dazzling midi voices in my older DOS based games. I missed the extra flares of music you could here in Mechwarrior 1's victory music (or bar scene music) with the primo soundcards.

Speaking of old sound cards, back in the 90's FM music synthesis seems to be the norm, with digitized sound effects thrown in courtesy of Yamaha chips, IIRC. Back then wavetable was considered high end.

These days FM music synthesis is pretty much obsolete. So what are we using these days? Wavetable? Does this mean today's onboard audio are actually wavetable-class audio devices? If yes, it's just awesome, isn't it?

PenGun wrote:

Waco wrote:

Generally if the output isn't noisy (the analog output that is) I don't really mind onboard audio.

My HTPC just uses the standard Realtek crap and since there's no audible hiss at the maximum gain I use on my receiver there's nothing to really complain about.

Funny, I prefer the sound from a good cassette deck, hiss and all, compared to many CD players and probably all of compressed audio. A Teac V8000S will murder most digital sound.

Can't tell how serious you are... cassette is notoriously limited in DR and consistency of sound. Most of the limitations associated with conventional "digital sound" - CD's - are due to mastering choices, not issues with technology.

ronch wrote:

Speaking of old sound cards, back in the 90's FM music synthesis seems to be the norm, with digitized sound effects thrown in courtesy of Yamaha chips, IIRC. Back then wavetable was considered high end.

One of the EE courses I took in college (back in the 1980s) covered FM synthesis. It was still considered fairly bleeding edge back then, since it provided a way to produce a wide array of sounds with (relatively) simple tech, and was much less resource intensive than wavetable or additive synthesis.

ronch wrote:

These days FM music synthesis is pretty much obsolete. So what are we using these days? Wavetable? Does this mean today's onboard audio are actually wavetable-class audio devices? If yes, it's just awesome, isn't it?

Yup, cheap DSPs pretty much killed FM synthesis, and cheap multi-core CPUs are in turn killing off the DSP-based soundcards. A modern CPU can do real-time wavetable synthesis in software without breaking a sweat.

I've been playing around a bit with music synthesis lately, using FluidSynth, Qsynth, Rosegarden, and JACK Rack. Not only can you do high-quality multi-channel wavetable synthesis, you can apply effects in real-time too (chorus, reverb, EQ, simulated guitar amps/cabinets, you name it). And since it is all done in software, it is completely soundcard agnostic.

Yeats wrote:

PenGun wrote:

Waco wrote:

Generally if the output isn't noisy (the analog output that is) I don't really mind onboard audio.

My HTPC just uses the standard Realtek crap and since there's no audible hiss at the maximum gain I use on my receiver there's nothing to really complain about.

Funny, I prefer the sound from a good cassette deck, hiss and all, compared to many CD players and probably all of compressed audio. A Teac V8000S will murder most digital sound.

Can't tell how serious you are... cassette is notoriously limited in DR and consistency of sound. Most of the limitations associated with conventional "digital sound" - CD's - are due to mastering choices, not issues with technology.

Have you any actual experience or are you just another, I read about this stuff and you are wrong, wienie?

PenGun wrote:

Have you any actual experience or are you just another, I read about this stuff and you are wrong, wienie?

First thing I did when digital came around was to buy CD copies of all the LPs I'd recorded to cassette during my college days. Even with a high-end deck (Denon DR-M44HX) and metal tapes (TDK MA-90) it was pretty obvious that cassette technology had run its race and lost. Maybe a Nakamichi Dragon or 1000ZXL might have changed my mind but I was never (or now) able to afford either.

PenGun wrote:

Have you any actual experience or are you just another, I read about this stuff and you are wrong, wienie?

I'd pit a good digitally mastered CD up against any tape deck you can imagine and expect the CD to win.

Yes, CDs aren't perfect, but to claim that a well-mastered CD is inferior to a tape simply because it's digital is hilarious. Do you really think recording studios use analog recording today? Do you really think the bitrate / frequency of a CD is low enough to cause it to be measurably inferior to a cassette tape?

PenGun wrote:

Yeats wrote:

Can't tell how serious you are... cassette is notoriously limited in DR and consistency of sound. Most of the limitations associated with conventional "digital sound" - CD's - are due to mastering choices, not issues with technology.

Have you any actual experience or are you just another, I read about this stuff and you are wrong, wienie?

Jeez, take a chill pill.

While cassettes could certainly provide audiophile-grade sound reproduction when using top-notch media and equipment and treated with care, they had a number of real-world practical limitations. Most people don't (and didn't, even back in the day) have $500 (let alone $1000) cassette decks capable of coaxing the best possible fidelity out of the format; prerecorded cassettes generally had abysmal fidelity (you had to dub them yourself from another high-quality source if you didn't want them to sound like utter crap); and they were prone to physical damage from heat, humidity, and dirty/malfunctioning playback equipment (the infamous "eaten tape" issue).

Many modern music CDs (and digital downloads) are indeed horribly (re)mastered, with way too much dynamic range compression. It squeezes the life out of the music. Some of my best sounding rips (in OGG format, I have a Sansa Clip+ and OGG seems to give me the best bang per bit) are actually from old vinyl I bought back in the 1970s and 1980s. (Yes, I know... blasphemy to the analog purists. But I'm not about to haul a turntable and LPs to the office to listen to music while I work.)

When I originally started ripping my old vinyl to digital a few years back, I was initially baffled by the fact that the perceived volume level of the rips was 6-10dB lower than contemporary recordings, even when the peak level was normalized to 100%. After doing a little reading, I realized that it is because most modern non-classical recordings have their dynamic range compressed by a similar amount.

Captain Ned wrote:

PenGun wrote:

Have you any actual experience or are you just another, I read about this stuff and you are wrong, wienie?

First thing I did when digital came around was to buy CD copies of all the LPs I'd recorded to cassette during my college days. Even with a high-end deck (Denon DR-M44HX) and metal tapes (TDK MA-90) it was pretty obvious that cassette technology had run its race and lost. Maybe a Nakamichi Dragon or 1000ZXL might have changed my mind but I was never (or now) able to afford either.

The Teac smoked both those puppies. Well there are differing opinions.

My point originally was that hiss does not determine sound quality.

just brew it! wrote:

PenGun wrote:

Yeats wrote:

Can't tell how serious you are... cassette is notoriously limited in DR and consistency of sound. Most of the limitations associated with conventional "digital sound" - CD's - are due to mastering choices, not issues with technology.

Have you any actual experience or are you just another, I read about this stuff and you are wrong, wienie?

Jeez, take a chill pill.

While cassettes could certainly provide audiophile-grade sound reproduction when using top-notch media and equipment and treated with care, they had a number of real-world practical limitations. Most people don't (and didn't, even back in the day) have $500 (let alone $1000) cassette decks capable of coaxing the best possible fidelity out of the format; prerecorded cassettes generally had abysmal fidelity (you had to dub them yourself from another high-quality source if you didn't want them to sound like utter crap); and they were prone to physical damage from heat, humidity, and dirty/malfunctioning playback equipment (the infamous "eaten tape" issue).

Many modern music CDs (and digital downloads) are indeed horribly (re)mastered, with way too much dynamic range compression. It squeezes the life out of the music. Some of my best sounding rips (in OGG format, I have a Sansa Clip+ and OGG seems to give me the best bang per bit) are actually from old vinyl I bought back in the 1970s and 1980s. (Yes, I know... blasphemy to the analog purists. But I'm not about to haul a turntable and LPs to the office to listen to music while I work.)

When I originally started ripping my old vinyl to digital a few years back, I was initially baffled by the fact that the perceived volume level of the rips was 6-10dB lower than contemporary recordings, even when the peak level was normalized to 100%. After doing a little reading, I realized that it is because most modern non-classical recordings have their dynamic range compressed by a similar amount.

As one who has Zappa's "Man From Utopia" as a CD, as the original record and as the later digitally remastered record I have some idea about 'mastering", among other things.

No I will not take any of your pitiful pills. I'm used to the high end stuff and people who have read some opinion or other and then use that as an argument as to fact will always get a similar response.

PenGun wrote:

My point originally was that hiss does not determine sound quality.

I'll grant you that. But (for most people at least) it is considered to be a fairly significant negative if it is present at obviously audible levels. Sometimes (e.g. older analog recordings) it is unavoidable. But there's no point in introducing it on purpose.

PenGun wrote:

My point originally was that hiss does not determine sound quality.

Of course not...but if I can hear hiss at normal listening levels then wouldn't you agree that it is a bad thing?

PenGun wrote:

Have you any actual experience or are you just another, I read about this stuff and you are wrong, wienie?

I didn't really say that you were wrong, just wasn't sure how serious you were about the sound quality of (most) cassettes, perhaps you were exaggerating to make a point?

I started on vinyl in the 70's, and also 8-track (yes), cassette - often with Dolby off, I'd rather hiss than the other unpleasantness Dolby NR often introduced - then CD/digital.

My point is, when discussing cassettes, there are obvious, physical limitations to the medium, and these limitations rear their heads no matter the equipment used.

JBI has already stated the primary issue with that's cropped up the last decade or two, horribly compressed DR, clipped peaks and valleys, killing the subtlety and nuances of the music... but that's due to human choices, not because it's "digital".

"Weenie", not "wienie".

Yeats wrote:

PenGun wrote:

Have you any actual experience or are you just another, I read about this stuff and you are wrong, wienie?

I didn't really say that you were wrong, just wasn't sure how serious you were about the sound quality of (most) cassettes, perhaps you were exaggerating to make a point?

I started on vinyl in the 70's, and also 8-track (yes), cassette - often with Dolby off, I'd rather hiss than the other unpleasantness Dolby NR often introduced - then CD/digital.

My point is, when discussing cassettes, there are obvious, physical limitations to the medium, and these limitations rear their heads no matter the equipment used.

JBI has already stated the primary issue with that's cropped up the last decade or two, horribly compressed DR, clipped peaks and valleys, killing the subtlety and nuances of the music... but that's due to human choices, not because it's "digital".

"Weenie", not "wienie".

I blame my spell checker.

The Teac was an unusual machine but my point is I'd rather listen to a little hiss on excellent sound than listen to a dead black background on crappy sound.

Yeats wrote:

I started on vinyl in the 70's, and also 8-track (yes), cassette - often with Dolby off, I'd rather hiss than the other unpleasantness Dolby NR often introduced - then CD/digital.

The main problem with Dolby NR was that it was extremely sensitive to variations in high frequency response. Any variation in the response at the high end would upset the Dolby tracking, resulting in audible artifacts in the treble range. Unless you used the exact media your deck was factory calibrated for, or had a deck with fine bias adjustment control and fiddled with it to find the right setting, this was a potential issue. And good luck if you played the tape back on something else (e.g. recorded on your home deck, but played back in your car or Walkman).

My today persona agrees completely. My 1979 persona wanted nothing more than to slot the cassette into the player, crank it up, and roll down the road at high volume and crappy sound. The '70's were a great time.

Well, we have been using a Realtek ALC889A for HTPC duties for years now. The onboard is responsible for D to A conversion to a Logitech z-5500 5.1 system.

Sounds great with blu-ray movies, music through the system is variable, with classical sounding the best. Haven't really compared it with anything else, because the family reckons it sounds awesome...

Dale

Captain Ned wrote:

And then we get into the "order" of the distortion, even or odd. Tubes distort in even-order, which sounds good to the ear (and is why really good guitar amps still use tubes) while sold-state distorts in odd-order, which sounds discordant to the ear.

Out of all the myths which exist about vacuum tubes and transistors this is by far the biggest. IIRC it dates back to a poorly written article published in a magazine in the mid-1970s.

rephlex wrote:

Captain Ned wrote:

And then we get into the "order" of the distortion, even or odd. Tubes distort in even-order, which sounds good to the ear (and is why really good guitar amps still use tubes) while sold-state distorts in odd-order, which sounds discordant to the ear.

Out of all the myths which exist about vacuum tubes and transistors this is by far the biggest. IIRC it dates back to a poorly written article published in a magazine in the mid-1970s.

Mmm... I wouldn't be quite so dismissive. There are a lot of factors involved. Tube amp output stages tend to be single-ended and transformer coupled instead of symmetrical (push-pull) and direct coupled, which will tend to reduce higher-order and odd harmonics relative to even lower order harmonics when the amp is overdriven. This will yield a less harsh, more musical character.

From an acoustic standpoint, it makes sense that even harmonics would be more "musical" -- the first two even harmonics (2x and 4x fundamental) are exactly one and two octaves above the fundamental. None of the odd harmonics have this simple octave relationship.

Sure, now that we understand this, we can design solid state amps with similar characteristics to a tube amp... especially when using FETs instead of BJTs, since FETs have electrical characteristics more closely approximating that of tubes. But that doesn't mean that "tube sound" is a complete myth.