Buying from HDTracks - optimal bitrate

Hi folks,

Here in SA we're subject to some crappy local CD pressings. Importing CDs via Amazon is an option, but gets pricey...and so, HDTracks starts to become competitive as an avenue to buy music.

Of course there's a proliferation of formats and bitrates...unless there's some unobtanium stellar recording on DSD I don't care at all, so regarding PCM: If you had a recording available in 44.1, 48, 88.2, 96, 176.4 or 192khz what would you elect to buy and why?

I have a GOV2, so can play any of these, but would also convert to 44.1 MP3 for portable listening and Bluetooth devices that won't support higher bitrates.

What's your thinking?

 

To be frank, I don't believe there is a lot of good definitive data on this topic. I could be wrong, but I haven't found any yet. I also don't believe many manufactures want to state anything and alienate their customers. Many private organizations aren't tackling this either. There are a few that I have found that make some good arguments, but like a lot of things on the Internet these days it is hard to find authoritative data that can be relied upon. Nonetheless, many newer DACs do support up to at least 192kHZ. Do you want to use a bit rate this high? Some of that comes down to the equipment you are using. For instance, as the bit rate goes up, some equipment is not going to be up to the task even if it states that it is supported. In addition, jitter needs to be increasingly better as your sample rate goes up. Some equipment has difficulty maintaining acceptable jitter levels even at 44.1kHz. Also, you need to consider the hard drive space requirements. Those high frequency files are going to take a lot of space. As a side note, age might come into play. As you approach your mid 40s, it is typical for most people to no longer hear frequencies above roughly 14kHz. If you believe the Nyquist Theorem is still valid (which music sampling rates have been based upon), you only need 2x the frequency to adequately describe a signal. Obviously, I'm glossing over things a little, you still have to consider other things such as low pass filters and etc. Nonetheless, if you can't hear over 14-15kHz anyway as you age, then even the regular 44.1kHz is still sampling at nearly 3x for the highest frequencies. In short, I'm not entirely convinced as you go up in sample rate you would get that much of a sonic benefit. In fact, things might be worse if your DAC is not up to the task (speed/clocking, jitter, and etc.), and you need a lot more hard drive space. Lastly, no matter how many samples you have, you still have samples and that will need to be reassembled back into an analog wave form. A lot of how good that will sound is based on the DAC's implementation, and I'm always amazed how good plain ol' 44.1kHz can sound on good equipment. Interestingly enough, bit depth has probably been a less controversial topic as of late, the main difference between say 16, 20, or 24 bits is pretty much just the noise floor. 16 bits is pretty good, and a higher number of bits might not gain you as much as you think. You really have to consider what the effective number of bits your DAC can really use. Without getting too technical, you might be surprised what a normal high-end "multibit" delta sigma DAC is capable of in regards to the effective number of bits... Also, I believe HDtracks does provide a lot its files via flac format (I haven't looked at them in a few years). Nonetheless, I use that format myself. If I remember correctly, the creator of that format (even though it supports many bit depths and bit rates) has some published articles on the virtues of 44.1/16 (redbook) format. In short, I believe that more is not always better in this case even if that might fly in the face of the prevailing "conventional wisdom" of more is always better. In other words, getting everything in 192/24 files might not give you the results you want depending on the equipment, interfaces (i.e. SPDIF, USB, optical, HDMI, or other), and etc. used. To make a long story longer, I have a tendency to believe that 44.1/16 is good enough for even high-end systems. If you were to consider anything more, I would increase the bit depth first before sample rate. I have not done any exhaustive testing on this myself even though I have an engineering background. This is because all my data files are 44.1/16 at this point (rips from CDs), and they are awesome even on higher end systems. In short, I have had no incentive to test alternative formats. But I have obviously devoted time to researching the merits of any hardware's ability to process these alternative formats.

 

Thanks for the reply...you're thinking along my lines that higher rates may come with unexpected disadvantages. There's potentially a world of devil in the details, like what the source is encoded in and so on that may not be known. I am perfectly happy with my vast majority of 44.1/16 CD rips so my question is really geared toward the pitfalls you describe.

I suppose file size is as good a yardstick as any, so one might do worse than simply getting the lowest rate available at the highest bit depth.

 

I typically don't spend the extra USD 7.00 to get 192/24 if 96/24 is available. Doesn't seem worth it to me; I'm much less confident of my ability to hear meaningful differences between 192/24 and 96/24 than between 96/24 and 44.1/16z

 

Another point to consider, when you purchase the highest bitrate (without going DSD) you are essentially getting 176 or 192khz in 24 bit. This frequency rate may be better suited for some gear that likes this rate instead of using algorithms to upsample to 176/192. A great example of this over sampling vs non oversampling would be the Schiit Bifrost Multibit that runs non oversampling at 176khz as well as 192khz. Is there a sound difference? I believe so. A bit to clarify, the Schiit Bifrost upsamples everything not 176khz or 192khz to one of those two sample rates depending on the lower sampling rate (44.1/88.2khz sample rates gets bumped to 176khz and 48/96khz gets bumped to 192khz)

Something about upsampling even with a perfect algorithm (by M Moffat in the Schiit units) still creates a certain sound that takes some natural aspecs out of the treble (to my ears). Hence for the Schiit Multibit Bifrost, I prefer 176khz and 192khz files. Even though I only own 3-4 albums in such rates and everything else is in redbook or 96khz for the sake of better mastering from HDtracks but not getting the extra $5 from me for that 192khz.

TBH I have a 44.1/48khz NOS 16bit dac I prefer over A LOT of these newer better dacs. I might eventually sell off the Bifrost Multibit.

 

"HIGH DEFINITION" DIGITAL AUDIO FOR LISTENING PURPOSES IS A SCAM!
Don't pay extra for it. The only way to hear the difference between a 16-bit and 24-bit master is to crank the recording to 140 decibels and hear the noisefloor. Redbook audio is good enough. There's no reason to even buy a master with a higher sampling rate than 44.1 or 48 kHz as mics don't capture ultra high frequencies reliably at all and you can't hear them. All that can happen is they bring noise into the audible band.

Learn how Nyquist-Shannon and PCM actually work. Consider buying import CDs from websites other than Amazon; Amazon jacks up the price ridiculously: $30 dollars for a 10 euro Dutch CD for example. CDs are lightweight, you can get them shipped anywhere for cheaper than Amazon's markup, and you can know your own rips of them are good as you use EAC. Most foreign CD pressings have identical masterings now too. Rip it securely with EAC and it will sound exactly the same. Who cares if the South African or American insert is slightly lower printing quality than the European digipack?

http://people.xiph.org/~xiphmont/demo/neil-young.html
https://wiki.xiph.org/Videos/A_Digital_Media_Primer_For_Geeks
https://wiki.xiph.org/Videos/Digital_Show_and_Tell
http://exactaudiocopy.de/
http://dr.loudness-war.info/ for checking masters.

 

I thought the whole point of the Schiit Multibit was that so CD Redbook music would sound it's best?

 

Really depends on the music though.

 

What does?

 

f**k, now I can't stand people questioning over and over again if going over 16 bits/44.1 khz will bring better results. The answer is simple. If you want the ultimate satisfaction of placebo and big dick syndrome, go for whatever bit rate that is highest, and it will instantly become amazeballs. Or go for DSD. DXD.

While listening blind, I'm not even sure if I can reliably discriminate 320kbps vs lossless, let alone asking me to pay for whatever higher than standard CD.

 

Well, if you were listening to poorly mastered CDs the Schiit multibit would make it sound worse because now you hear everything.

Back on track: A perfect example of this HDtracks having better sound quality over CD would be their remasters. The difference between RHCP Stadium Arcadium for example (a CD I played over and over until the laser wore out on my CD player) is night and day between HD tracks and the original 2006 release. Even though there shouldn't be a need for the higher rates and arguably the 24-bit depth, the extra $$ or re purchase price is warranted to for such albums and the remaster. However not all albums will sound better, I find most Chesky albums as good in redbook as it is on 192khz 24bit.

 

I'm not ready to go "back on track" yet. I have seen a number of contrary impressions on this forum re the "worsening" of bad music through the Schiit multibit gear. If you have first-hand experience with Schiit multibits, then great; but so far your statements seem very conjectural.

 

Which part didn't you get? I find the Bifrost Multibit to sound better when it's not oversampling files to 176/192 because the source files from HDtracks are already 176/192. Feeding it with lower sample rates creates an oversampling effect on the treble. That's one aspect that I find HDtracks to be superior (do I need to put YMMV?). And of course the remastering I mentioned afterwards.

 

None so blind as those that rip out their own eyeballs.

 

Unfortunately, I think the thread got side tracked a bit. The OP wanted to determine whether higher bit rates or bit depths are better or worth it. Sure you will find some people that state they can hear a difference, and there is no wrong answer in regards to preferences. Equipment can surely have an effect on what is heard as well. But the thread turned into a debate of the performance of the Bifrost Multibit, and I will have to respectively disagree with the dissenting opinion on that point. It was mentioned that the higher bit rates sound different particularly with the Bifrost, and that might be. Obviously, the superburrito filter is not applied to 176.4 kHz or 192 kHz bit rates, and I don't believe the Bifrost has nearly the same level of reclocking that the Gungnir or Yggdrasil has. Due to this, the source is probably going to have a bigger impact on the sound quality because it will need to be extremely precise to not cause jitter issues that doesn't effect what is heard. Nonetheless, based on my own testing and performance testing/measurements and impressions of the Bifrost Multibit on this forum, I would say stating that the Bifrost Multibit as not accurate or natural sounding would be a mischaracterization. This is not to defend this product because I have another DAC, the Violectric V800, which is used in some studio applications, and I found it less natural and accurate sounding than the Bifrost. So even though I have another DAC that is found in some studio applications and is much more expensive than the Bifrost. I'm not willing to say it is better than the Bifrost in the least. Back to what the OP wanted to know. From a bit rate or depth perspective, I don't believe the science and technology of today supports using higher than normal 44.1/16 or even 44.1/24 that will have an audible difference to your benefit. In addition, just like buying equipment there is a diminishing return even if there was. Also, just because music is delivered in higher bit rates doesn't mean that it was mastered well either. In fact, you can tell that a lot of sound engineers don't use headphones to assist with their mixing. If they did, they would catch a lot of their mistakes such as clipping and etc. that is easily heard over headphones which can get missed when just listening to studio monitor speakers alone. In short, my advise is save your money and buy 44.1/16 or 44.1/24 music which will have the most compatibility with any equipment that you are likely to encounter for the foreseeable future. Keep in mind that us audio nuts typically prefer higher fidelity than most, and even at these rates they are significantly better than what you typically find the general population using with mp3 music files and the like. For the longest time, I could not even listen to mp3s. It has only been within the last couple of years that algorithms and even at the highest settings for encoding that I am able to enjoy music reproduced from mp3s which I still only use in portable audio.

 

After purchasing 25 albums in various "HD" bit depths/sample rates up to dsd64, I have concluded that with my ears/gear, it is not worth it to rebuy, at $18-25, music that I have already bought at least once and in many cases 3 times. One thing I do know however, is that music that is well recorded and mastered sounds better in redbook than much music, especially modern recordings, in "HD." Of couse, the bigger picture is that i'd rather listen to music that I like in mp3 form than music I don't like in dsd256 or 24/192.

 

To answer the OP's original question, the optimal bitrate depends on the provenance of the album. Some albums were recorded or mixed digitally at a time when 16/44 was the best you could get, so there's no real benefit to going over 16/44. A good example of this is Paul McCartney's Pipes of Peace, which he had to remix in order to get a higher res version, because the original masters are locked at 16/44. In other cases, like the recent Led Zeppelin remasters, Jimmy Page opted for 24/96. You should get those remasters at 24/96 as a result.

The problem here is that a lot of releases don't give us detailed provenance, so it can be a guessing game. And HD Tracks has been caught upsampling before, so you can't assume that because a 24/192 version exists, that it actually makes use of 24 bits and 192khz sample rate. It can take a bit of digging, and sometimes you just have to take the leap.

As for whether or not one can hear differences in 16 and 24 bit files, here is what I did. I took several sample tracks of various songs I knew well, some were rock/pop, some were classical. I got 24/192 files that I knew weren't upsampled, as well as some DSD64 and DSD128 files that were natively recorded in those formats. I then downsampled them myself into 24/96 and 16/44. For the DSD files, I kept the originals, made new FLAC copies, and then did the same. I played them through my Pono player, using HD600's in balanced mode, and tried to see if I could hear any differences. I spent many hours going back on forth, listening to snippets of these songs. Every time I thought I detected something unique in the 24 bit files, I'd switch back to the 16, and lo and behold, the same sound was there too. I personally didn't feel I was missing any information.

I plan to do this test again when I upgrade my DAC to a multibit. Perhaps the additional realism and resolution will make the differences more apparent to me. Of course, my test wasn't wholly scientific, but I wasn't trying to prove anything except whether or not I felt I could hear a difference, and I simply could not. The same goes for comparing DSD to FLAC. The files sounded the same to me, at all resolutions.

So for my current system, I don't bother with 24 bit. 16 bit should be able to hold all the audible information we can perceive. I've heard many arguments as to why 24 bit is better. This may be confirmation bias, or it may be people with more sensitive hearing. It doesn't really matter. All that matters is what YOU can hear so you can decide for yourself what is the optimal bitrate.

 

Love this

I've heard many arguments as to why 24 bit is better. This may be confirmation bias, or it may be people with more sensitive hearing. It doesn't really matter. All that matters is what YOU can hear so you can decide for yourself what is the optimal bitrate.

 

Thanks! It took me a while to do it all, and every time I thought I had a eureka moment with 24 bit, the 16 bit fought back and proved me wrong.

Also, it was nice to be able to compare native DSD against FLAC, since the Pono can do both. I now have a GOV2, which can also do both, but I don't bother with DSD because there are very few albums recorded natively in DSD, which is the only reason I'd care.

 

Personally, I research every purchase I make at HD Tracks as there are way too many shenanigans going on there to buy blind. The computer audiophile site is a good resource as they will post impressions, a lot of times measurement backed, that give you a better idea if the higher sampling/bitrate translate to any meaningful difference in the source material. Now whether your gear can handle it or your ears can perceive it is of course still up for debate.