"Native" Hi-rez vs upsampled Hi-rez

Apologies in advance. I'm sure this has been discussed before but my feeble searching skills didn't turn up anything.

As I understand it, modern digital music is mastered at 192 kHz and then downsampled to 44.1 for CD, etc. I haven't been able to find information about Hi-rez music files you buy as FLACs, but my hope would be that they were sold at their "native" resolution and not messed with in between.

Assuming that I haven't gotten this totally messed up, has anyone compared "native" hi-rez to upsampling from 44.1? I've been a 44.1 Redbook guy, mostly due to the limited selection of hi-rez available. When I have listened to sample tracks at different resolutions, I can hear the difference, but it hasn't been a worthwhile difference to me. But there has been so much development in upsampling and enthusiasm for it that I find myself looking at it again.

I'm also curious, if you like hi-rez music, why don't you just upsample and save the hi-rez file instead of converting it as you play it every time? After all, storage is cheap and you could store a very large library of hi-rez files.

 

I think it depends a lot on who's recording and mastering it, you'll find stuff's masters being stored as DSD from larger music production companies vs something closer to what you're imagining from smaller recording studios or DIY operations and I'm sure all sorts of other variations in-between that I'm not familiar with - stuff is absolutely encoded differently for different release intentions and I think it's pretty rare that you'd find anything that hasn't been "messed with" since it usually just doesn't make sense to release things at that resolution without getting into niche territories like SACD etc.

"native" hi-rez files (should) contain data for higher frequencies than an oversampled track does, going from 192khz to 44.1khz for example is a lossy operation and you can't just get that data back by oversampling it, you can take a look at spectrograms of your music files to see that (you can also usually tell via spectrogram if something was a lossy master that was released as lossless) - the point of interpolation (oversampling) for playback is different from the point of native high resolution files, I'm sure there are threads on here like the HQPlayer thread or NOS (non over-sampling) dac threads where you could find out more about playback oversampling

I think some people do in fact do what you describe in some edge scenarios like converting to DSD or some other fancy/hard to compute end state, but generally it's just not that compute intensive to do on the fly with current technology so there's no point to storing it that way - there are also different methods/algorithms for doing the oversampling and people have a lot of different tastes/opinions there. the schiit dacs for example have their own proprietary filter for doing their internal oversampling and you'll see it on some newer stuff where there are different algorithm selections for the dac oversampling - I think the wavedream for example has a couple different options here

 

Thanks, decomposition! You answered my questions clearly and I learned something. I think I will just not worry about it for a while longer.

 

Maybe you find this free download interessting.

https://soundliaison.com/products/compare-formats-free-downloads

Which format is best for your DAC?
Every DAC has it's own character, some sounds better using PCM files and some sounds better using DSD files......some people find the convenience of FLAC files more important.
Here on this page is a list containing a song in 10 different formats.

I especially lIke this album from them:

Feenbrothers - Play Dave Brubeck

https://soundliaison.com/products/feenbrothers-play-dave-brubeck

 

From what I understand from my experience upsampling using HQ player is that the filter math works better and becomes more flexible at higher resolutions. With some DACSs, if you upsample PCM far enough, you can bypass their internal filters and define the filtering in software. Take this with a grain of salt, though, as I understand this process in the most basic way.

What I can say with more confidence it that upressing to DSD using HQplayer and a DAC that has DSD-Direct (which bypasses the DAC's filters and conversions) changes things. It allows me to pick and choose the filters and modulators to suit the material and my chain at the time. The difference can be subtle or dramatic...but it's always enough of a difference that I get a better listening session. There are several dozens of possible combinations with HQ Player, I've settled on a small handful that I like.

It's a fiddly process, though. Sometime you just want a DAC that will chef it up for you so you can get to listening to the music. (Looking at you Yggdrasil LIM)

 

Mastering over caring about hi rez. Even the 90s Steve Hoffman mastered stuff from the 90s sounded a lot better to me than most hi rez remastering.

 

I think this is an apt metaphor. Though what follows is a bit of a stretch of "chefing it up", but if you can't name all the stations in a brigade de cuisine, or even name the 5 mother sauces, you might have a difficult time improving on a well prepared 5-course meal at a nice restaurant, if trying to prepare it at home (not saying anyone can't, but for the un-chefy, there's gonna be a lot of burnt and poorly seasoned food). likewise, dicking around with the knobs in HQPlayer won't net you much if you haven't read something like this: https://www.mathworks.com/help/signal/ug/practical-introduction-to-digital-filter-design.html

(which i've read, and reread multiple times and still come back to LIM as being more "right" than anything I came up with in HQPlayer on my own). But I'm a neanderthal, so...

 

No argument here. I've been working with HQ Player for 5 or 6 years and still get lost. I have come up with my favourite combinations by talking with the developer, Jussi, trial and error, and following the massive thread at computer audiophile, not through any deep understanding of how it all works. I know I prefer linear phase long filters rather than minimum phase or short filters. I keep going back to his poly-sinc-xtr-lp filter despite there being newer developments.

I appreciate the article - I have been meaning to some more research on this. I'll dig in.

 

Most DACs will upsample, and they do it in a way that works with their DSP chips. Especially in cheaper DACs, those don’t provide lots of processing power, so the algorithms used are optimised for speed, not for best possible sound. That’s why upsampling on a PC can yield much better results, especially with a product like HQPlayer with all its possibilities and attention to the smallest details.

Personally, I didn’t hear much of a difference with Roon upsampling with my Meier Daccord FF DAC, minimum phase sounded the same, and linear phase a tiny bit different. The RU6 I had at the time, however, sounded quite a bit better to my ears when fed with Roon upsampled material, so it does depend a lot on what the DAC does.

HQPlayer is quite a large step up from Roon upsampling on my Daccord DAC, though my guess is that a good clock is required to really make it worthwhile (in the DAC when using USB, in the streamer for most DACs when using optical, coax, or AES connections): With the RU6, the difference wasn’t as striking by far.

Anyway, about the actual questions: To my ears, material that comes in hi-res is more or less equivalent in quality to upsampled material. There are small differences in favour of hi-res material, but really small. I’ve seen or rather heard some cases (very few) where the HQPlayer upsampled version seemed better to me (don’t remember what that was, unfortunately, some classical stuff, I think). As there are rumours that some hi-res material is actually just upsampled by the provider, those might be such cases? In general, though, hi-res wins over upsampling.

Of course, it would be possible to upsample everything and save it that way. There is software out there that is meant to be used like that, doesn’t support upsampling on the fly. Especially with HQPlayer, I wouldn’t do that, though, because when I buy a new DAC, or even when just my preferences change, I might want to use different algorithms. I’ve still thought about that, though, because there’s no good way to feed my Dynaudio Music 5 directly from HQPlayer (digital input only up to 96 kHz, but can play 384 kHz via UPNP). But in the end, it’s too much hassle.