Jump to content
IGNORED

Morals of Upsampling


Recommended Posts

Ok, so here goes: in our endless pursuit of best sound and music truth, I have started wondering whether or not upsampling is moral or not. Hear me out, as we all try and spend hundreds/thousands of bucks on hi-end equipment in order to be able to experience first-hand that particular concert or make it seem as we are in the studio at that recording session.

 

Yet, lately, I see a lot of people upsampling everything (either PCM or to DSD), in an effort to make the sound sweeter or less edged or more analogue-like, etc. So my concern is, how can we call ourselves audiophiles and "seekers of the music truth" if we start covering up these sound problems that may exist, in an effort to make the music fit to our ears and personal preferences?

 

Are we in the end some form of "hypocrites"? I am asking this, because there are great recordings and then there are average recordings or even lower-quality recordings (because of the tech used or the studio equipment or the era they were recorded or the mind of the engineer and/or the producer, etc).

 

At the end of the day, are we truly searching for the truth or are we trying to find ways to make all recordings suit our personal tastes and ears?

 

The reason I am asking this is because with upsampling, new information-data is introduced and so far all the comments I've read are how better the music sounds from the original (especially with DSD upsampling). Also, forgive me, I am not a tech-guru nor an upsampling master engineer.

 

Please, I am not criticising anyone nor do I want to start a war, it's just an honest thought I've been having lately, especially with all the DSD upsampling I have been reading lately.

 

In the end, my opinion is that a "rough" recording needs to stay "rough", because that's the "real" sound for whatever x/y reason this had happened. Am I overthinking this? Or am I thinking wrongly that audiophilia is about the musical truth in each recording, when in essence it may be something else?

 

 

Sent from my iPhone using Computer Audiophile mobile app

Link to comment
Share on other sites

I don't think it is any more immoral than reaching for the volume control knob. The way I look at it, the goal is to achieve bit-perfect playback as a starting point. Then, if you want to do DSP room correction, apply DAC filters, etc., that is entirely your choice as an individual. What bothers me from a "moral" perspective is having someone else force such choices on you (eg: dynamic compression/loudness).

 

Unless you are listening near-field with perfect speakers, the room and everything else is causing what you hear to differ from "the truth" or the true, unaltered playback.

Link to comment
Share on other sites

Yet, lately, I see a lot of people upsampling everything (either PCM or to DSD), in an effort to make the sound sweeter or less edged or more analogue-like, etc. So my concern is, how can we call ourselves audiophiles and "seekers of the music truth" if we start covering up these sound problems that may exist, in an effort to make the music fit to our ears and personal preferences?

 

I think you are misunderstanding why people upsample. The reason to upsample is a) to reduce the workload in your DAC and feed it information at the rate to which it would otherwise upsample itself (almost all DACs upsample themselves, so you aren't doing anything that wasn't going to happen if you just fed your DAC 16/444 input material) and b) to be able to use better/different filters in the process. Admittedly, in choosing a filter we may be choosing a different set of priorities than the original recording engineer; but then, he/she probably wasn't listening to the recording on your DAC when they were making those filter choices.

Synology NAS>i7-6700/32GB/NVIDIA QUADRO P4000 Win10>Qobuz+Tidal>Roon>HQPlayer>DSD512> Fiber Switch>Ultrarendu (NAA)>SMSL M500 DAC> Bryston SP3 pre>Levinson No. 432 amps>Magnepan (MG20.1x2, CCR and MMC2x6)

Link to comment
Share on other sites

Ah ... as long as you don't up sample x4 and then divide the file using every 1/4th sample and then sell the 3 new files ...

 

But if you are just trying to make the sound better that falls under: "pursuit of happiness" which, at least in the United States, is considered a basic right: https://en.m.wikipedia.org/wiki/Life,_Liberty_and_the_pursuit_of_Happiness.

 

That way of looking at the world is not universal, and clearly there are cultures where the pursuit of musical bliss is considered immoral (can you say ISIS?) There is also the modern Russian adage "a person who smiles a lot is either a fool or an American" https://hbr.org/2012/01/the-history-of-happiness

Custom room treatments for headphone users.

Link to comment
Share on other sites

Once you realize it's not about adding new information but about better filtering, the choice becomes where you want this to occur, if at all, in the software or the hardware. Software gives us great choice and experimenting with different filters and values can be fun. Which sounds best is quite personal. For inline up-sampling HQPlayer, JRMC, and Audirvana + all do a great job. Keep in mind it depends on the DAC. I have 3 DACs for example, which IMO sound better without prior up-sampling.

 

So, no morality crisis at all because we are indeed trying to extract as much as we can from the recording, warts and all.

 

You might enjoy this short article from 2003. It explains the reasons for up-sampling and over-sampling quite well, even for us non-techie types: SoundStage! Getting Technical Back-Issue Article (11/2003)

Link to comment
Share on other sites

The reason I am asking this is because with upsampling, new information-data is introduced

 

With upsampling no new information is introduced but better reconstruction of the original analog signal can be achieved with our real (non ideal) DAC chips.

 

The only information input of upsampling is the original lower sample rate recording. No more information than that is available about original analog signal, regardless on further processing of that information. It is like making a higher resolution picture from a lower resolution picture.

Link to comment
Share on other sites

Ah ... as long as you don't up sample x4 and then divide the file using every 1/4th sample and then sell the 3 new files ...

 

But if you are just trying to make the sound better that falls under: "pursuit of happiness" which, at least in the United States, is considered a basic right: https://en.m.wikipedia.org/wiki/Life,_Liberty_and_the_pursuit_of_Happiness.

 

That way of looking at the world is not universal, and clearly there are cultures where the pursuit of musical bliss is considered immoral (can you say ISIS?) There is also the modern Russian adage "a person who smiles a lot is either a fool or an American" https://hbr.org/2012/01/the-history-of-happiness

+ 1

I like your sense of humor :)

Alain

Link to comment
Share on other sites

Once you realize it's not about adding new information but about better filtering, the choice becomes where you want this to occur, if at all, in the software or the hardware. Software gives us great choice and experimenting with different filters and values can be fun. Which sounds best is quite personal. For inline up-sampling HQPlayer, JRMC, and Audirvana + all do a great job. Keep in mind it depends on the DAC. I have 3 DACs for example, which IMO sound better without prior up-sampling.

 

So, no morality crisis at all because we are indeed trying to extract as much as we can from the recording, warts and all.

 

You might enjoy this short article from 2003. It explains the reasons for up-sampling and over-sampling quite well, even for us non-techie types: SoundStage! Getting Technical Back-Issue Article (11/2003)

 

Thank you Melvin, I'll have a read. Something tells me that I have misunderstood the whole concept of upsampling, so I'll get back to research!

 

 

Sent from my iPhone using Computer Audiophile mobile app

Link to comment
Share on other sites

The reason to upsample is a) to reduce the workload in your DAC and feed it information at the rate to which it would otherwise upsample itself

 

+1

DIY coax tractrix horn system 2 corner subwoofer /// 6 full digital amplifier D802 floating PSU 12V battery & caps/filter /// Active crossover @ Acourate Convolver & room [email protected] /// General 2 PC setup: floating PSU picoless battery & caps/filter powered Bicker DC160W: PC1(Player) - individual stripped MS RamOS with JPlay /// PC2: Server 2016 RamOS - AO 2.0b5 - Acourate Convolver - online convolving & crossover /// Chain: PC1 - USB - F-1 - SPdif coax - Mutec MC3.1+ USB - SPdif coax - FireFace UCX floating PSU 12V battery & caps/filter - USB - PC2 - FireFace UCX - Adat LWL - Mutec MC-4 - 3 x SPdif coax - D802 Low/ D802 Middle / D802 High - 2 Stereo Lab KWH250 with BMS 4590 plus 2 modified corner subwoofer Abacus ABS210

Link to comment
Share on other sites

+1

 

So here's how hypocrite I am, lol!! I actually got an iFi Nano DSD the other day. I use Daphile OS and the question is: should I upsample PCM to DSD? What are your personal suggestions?

 

 

Sent from my iPhone using Computer Audiophile mobile app

Link to comment
Share on other sites

I actually got an iFi Nano DSD the other day. I use Daphile OS and the question is: should I upsample PCM to DSD? What are your personal suggestions?

Hi pipis2010,

 

if if you have everything at hand: Try it, listen, listen, listen and make up your mind ;) I have no experience with the iFi Nano DSD.

 

Take care

Thomas

DIY coax tractrix horn system 2 corner subwoofer /// 6 full digital amplifier D802 floating PSU 12V battery & caps/filter /// Active crossover @ Acourate Convolver & room [email protected] /// General 2 PC setup: floating PSU picoless battery & caps/filter powered Bicker DC160W: PC1(Player) - individual stripped MS RamOS with JPlay /// PC2: Server 2016 RamOS - AO 2.0b5 - Acourate Convolver - online convolving & crossover /// Chain: PC1 - USB - F-1 - SPdif coax - Mutec MC3.1+ USB - SPdif coax - FireFace UCX floating PSU 12V battery & caps/filter - USB - PC2 - FireFace UCX - Adat LWL - Mutec MC-4 - 3 x SPdif coax - D802 Low/ D802 Middle / D802 High - 2 Stereo Lab KWH250 with BMS 4590 plus 2 modified corner subwoofer Abacus ABS210

Link to comment
Share on other sites

Hi pipis2010,

 

if if you have everything at hand: Try it, listen, listen, listen and make up your mind ;) I have no experience with the iFi Nano DSD.

 

Take care

Thomas

 

On another note, through Daphile I can have the option to play DSD natively or via DoP. I tried to upsample 24/192 PCM to DSD and via native I get a weird noise and after a while the music stops. Anyone knows why this happens? My cpu is Athlon 5350, 2.05 GHz with 4GB RAM DDr-3.

On the other hand the same upsampling procedure via DoP seems to work fine.

 

Am I missing something?

 

 

Sent from my iPhone using Computer Audiophile mobile app

Link to comment
Share on other sites

Thank you Melvin, I'll have a read. Something tells me that I have misunderstood the whole concept of upsampling, so I'll get back to research!

 

I believe many of us were in the same boat really. Fortunately, some very smart and very patient members have over time enlightened us with the concept, helping us understand a bit more about the inner working of DACs.

 

Have fun with the new Nano :)

Link to comment
Share on other sites

I am not usually the one that resorts to bold statements but in this instance, I'll make an exception: up-sampling to DSD has saved my life! Plain and simple! Well. that and the Lampizator... Finally, I am listening to music again and I mean MUSIC, not just frequencies or sounds or whatever... Up-sampling done right does not make the music softer or by any means embellished but rather the opposite, as transients become clearer, timbre more clearly differentiated and the natural ebb and flow in the music is released from its constraints.

I'll be back in Athens in late February. If you're anywhere near that, come and have a listen... Listening is believing! (oops! another bold statement...)

Ted

Link to comment
Share on other sites

I am not usually the one that resorts to bold statements but in this instance, I'll make an exception: up-sampling to DSD has saved my life! Plain and simple! Well. that and the Lampizator... Finally, I am listening to music again and I mean MUSIC, not just frequencies or sounds or whatever... Up-sampling done right does not make the music softer or by any means embellished but rather the opposite, as transients become clearer, timbre more clearly differentiated and the natural ebb and flow in the music is released from its constraints.

I'll be back in Athens in late February. If you're anywhere near that, come and have a listen... Listening is believing! (oops! another bold statement...)

Ted

 

I actually live in Athens! Send me a pm with your contact info, cause I would love to listen to that Lampi sometime. I envy you btw :))

 

 

Sent from my iPhone using Computer Audiophile mobile app

Link to comment
Share on other sites

So here's how hypocrite I am, lol!! I actually got an iFi Nano DSD the other day. I use Daphile OS and the question is: should I upsample PCM to DSD? What are your personal suggestions?

 

Try and see what you like. With PCM input, the iFi Nano offers a choice of two filters for its internal upsampling. Note, however, that although the switch positions are labelled "standard" and "minimum phase", both settings are actually linear phase. The difference is that the "minimum phase" setting has a slow roll-off which reduces the ringing a bit. There is also a third, less obvious, option: if you upsample to 352/384 kHz in software, the internal filters in the DAC chip (TI/BB DSD1793) are bypassed completely, so you might want to try that as well.

Link to comment
Share on other sites

On another note, through Daphile I can have the option to play DSD natively or via DoP. I tried to upsample 24/192 PCM to DSD and via native I get a weird noise and after a while the music stops. Anyone knows why this happens?

 

Can you describe the noise? Do you get a) proper music for a while followed by noise, b) music with overlaid noise, or c) only noise? Does it depend at all on the file being played?

 

My cpu is Athlon 5350, 2.05 GHz with 4GB RAM DDr-3.

On the other hand the same upsampling procedure via DoP seems to work fine.

 

That's curious. The DSD modulator can get into a bad state with some evil inputs (full-range square waves and such). If some actual music is causing the modulator to destabilise, I'd very much like to know (I wrote it). The fact that it works with DoP suggests this isn't the problem though, since everything up to and including the DSD modulator is the same in both cases. After the modulator, the DoP path is probably slightly less demanding of the CPU, so that could be something. Although the main CPU hog is the modulator itself, it is possible that the difference is just enough to push it over the edge in one case.

Link to comment
Share on other sites

Can you describe the noise? Do you get a) proper music for a while followed by noise, b) music with overlaid noise, or c) only noise? Does it depend at all on the file being played?

 

 

 

That's curious. The DSD modulator can get into a bad state with some evil inputs (full-range square waves and such). If some actual music is causing the modulator to destabilise, I'd very much like to know (I wrote it). The fact that it works with DoP suggests this isn't the problem though, since everything up to and including the DSD modulator is the same in both cases. After the modulator, the DoP path is probably slightly less demanding of the CPU, so that could be something. Although the main CPU hog is the modulator itself, it is possible that the difference is just enough to push it over the edge in one case.

 

 

Hello mansr,

 

Concerning your first questions about the noise, I made some changes on the Advanced Player Settings in the Daphile settings and it seems to have worked. Nonetheless, the noise I was getting constantly throughout the playing of the song was this vinyl scratch-like noise and after a few seconds dropouts began to happen, then the music would stop and then it would continue again with dropouts (or it gave you the sense that it will start dropping out).

 

Moving on to the second set of questions regarding DSD modulation, please check the 2 screenshots I got from the Advanced Player Settings in Daphile:

 

Advanced Settings - Native DSD.JPG

 

I actually got the Native DSD playback to work without any hiccups or dropouts, by doing the following changes (which you can't see on the screenshots):

 

1) I changed the "Phase response" from "Minimal" (which probably caused the whole mess to "Linear Phase" (as per your suggestions for the 2 on-board filters that the iFi DSD has.

 

2) The "Attenuation" I changed it back to the default "-1 dB" value (previously it was set at -2 dB.

 

3) In the track transition delays area, I changed both settings to "0ms" (from the "100 ms" I had previously selected).

 

That's all the changes I made.

 

Now, everything seems to work and I have tested the above settings with FLAC files of 24/96 and 24/192 and 24/88.2 and they all work marvelously, but I do have one last and very important issue/problem: Every time, I click on next/previous track, or every time I select a different album or a different song to play, I get this loud instantaneous "bang" - it's the exact same noise that we get, when we try to connect a component on an open amplifier and once we start connecting each of the L/R cables it makes this loud bang. I get this loud, instantaneous bang EVERY time I select another track or another album and even during the transition from one track to another (even if I leave the tracks to playback gapless). If I press Play, then Pause and then Play again on a song, it doesn't make this sound-bang.

 

Any ideas on this? I am afraid that it will eventually hurt my speakers and that is why I think I will change back to the "DoP" DSD Playback setting in Daphile (with which I had no problems at all).

 

Just FYI, I am using a DIY Athlon 5350 (2.05 GHz) CPU, 4GB RAM DDR-3, 2.5" External HDD connected via USB 3 port with all my media files, a Paul Pang USB v2 card with an iFi LPS, an Audioquest Jitterbug and a cheap USB cable (actually the short USB cable that came in the box with the Nano DSD unit) between the PPA card and the iFi DSD unit. I don't know if any of these will help you figure out anything more...

 

Lastly, when the "Native" setting in DSD Playback works, IT IS PURE MAGIC!!!! My respect to all the people who upsample to DSD!!! Seriously, guys, this is pure audio N-I-R-V-A-N-A!!!!

PCM-DSD Upsampling - Native DSD.JPG

Link to comment
Share on other sites

In the end, my opinion is that a "rough" recording needs to stay "rough", because that's the "real" sound for whatever x/y reason this had happened. Am I overthinking this? Or am I thinking wrongly that audiophilia is about the musical truth in each recording, when in essence it may be something else?

 

If technical methods, like upsampling, give possibility to improve sound quality in complex with hardware (see more here http://www.computeraudiophile.com/f11-software/offline-upsampling-20999/index15.html#post512774), why not use upsampling?

 

Record can't stay "rough". In DAC it will upsampled almost in 100% of cases.

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment
Share on other sites

Hello mansr,

 

Concerning your first questions about the noise, I made some changes on the Advanced Player Settings in the Daphile settings and it seems to have worked. Nonetheless, the noise I was getting constantly throughout the playing of the song was this vinyl scratch-like noise and after a few seconds dropouts began to happen, then the music would stop and then it would continue again with dropouts (or it gave you the sense that it will start dropping out).

 

This sounds like your CPU isn't quite keeping up so you're getting buffer underruns of varying duration.

 

I actually got the Native DSD playback to work without any hiccups or dropouts, by doing the following changes (which you can't see on the screenshots):

 

1) I changed the "Phase response" from "Minimal" (which probably caused the whole mess to "Linear Phase" (as per your suggestions for the 2 on-board filters that the iFi DSD has.

 

2) The "Attenuation" I changed it back to the default "-1 dB" value (previously it was set at -2 dB.

 

3) In the track transition delays area, I changed both settings to "0ms" (from the "100 ms" I had previously selected).

 

That's all the changes I made.

 

Strange. None of those settings ought to make a difference to the CPU load. If you feel like experimenting, it would be interesting to see if you can reproduce the problem with just one of them set to the "bad" value.

 

Now, everything seems to work and I have tested the above settings with FLAC files of 24/96 and 24/192 and 24/88.2 and they all work marvelously, but I do have one last and very important issue/problem: Every time, I click on next/previous track, or every time I select a different album or a different song to play, I get this loud instantaneous "bang" - it's the exact same noise that we get, when we try to connect a component on an open amplifier and once we start connecting each of the L/R cables it makes this loud bang. I get this loud, instantaneous bang EVERY time I select another track or another album and even during the transition from one track to another (even if I leave the tracks to playback gapless). If I press Play, then Pause and then Play again on a song, it doesn't make this sound-bang.

 

Any ideas on this? I am afraid that it will eventually hurt my speakers and that is why I think I will change back to the "DoP" DSD Playback setting in Daphile (with which I had no problems at all).

 

Transitions are are tricky with DSD. Most likely, some buffer is being padded with zeros rather than DSD silence. If the DoP mode works better, I suggest using that. There is absolutely no difference to the sound while it's playing.

 

Just FYI, I am using a DIY Athlon 5350 (2.05 GHz) CPU, 4GB RAM DDR-3, 2.5" External HDD connected via USB 3 port with all my media files, a Paul Pang USB v2 card with an iFi LPS, an Audioquest Jitterbug and a cheap USB cable (actually the short USB cable that came in the box with the Nano DSD unit) between the PPA card and the iFi DSD unit. I don't know if any of these will help you figure out anything more...

 

I'm not familiar with that CPU, but it doesn't seem to be exceptionally fast. It does have the AVX instruction set though, but I don't think Daphile is built to make use of this.

Link to comment
Share on other sites

Can you tell/direct us to more information on your PCM>DSD conversion? I just now noticed its inclusion into Daphile.

This sounds like your CPU isn't quite keeping up so you're getting buffer underruns of varying duration.

 

 

 

Strange. None of those settings ought to make a difference to the CPU load. If you feel like experimenting, it would be interesting to see if you can reproduce the problem with just one of them set to the "bad" value.

 

 

 

Transitions are are tricky with DSD. Most likely, some buffer is being padded with zeros rather than DSD silence. If the DoP mode works better, I suggest using that. There is absolutely no difference to the sound while it's playing.

 

 

 

I'm not familiar with that CPU, but it doesn't seem to be exceptionally fast. It does have the AVX instruction set though, but I don't think Daphile is built to make use of this.

Forrest:

Win10 i9 9900KS/GTX1060 HQPlayer4>Win10 NAA

DSD>Pavel's DSC2.6>Bent Audio TAP>

Parasound JC1>"Naked" Quad ESL63/Tannoy PS350B subs<100Hz

Link to comment
Share on other sites

+ 1

I like your sense of humor :)

 

...and another + 1.....well played.

MacPro 24 GB/8TB / A+ and Pure Music / LAMPIZATOR B7 as Dac and Pre w/volume control / Regen /Intona USB Iso/curious USB / Emotive Audio Sira / PASS Aleph P / D-Sonic M3 600 mono's / MIT / Shunyata Alpha HC, Alpha Digital, Alpha Analogue, Weizhi PR-6 / Exact Power XP15a-4 balanced power distributor / FOCAL Maestro UTOPIA III's in red/black

 

"In life's final analysis, one's relationship with his Maker is all that really matters."

Link to comment
Share on other sites

Can you tell/direct us to more information on your PCM>DSD conversion? I just now noticed its inclusion into Daphile.

 

The conversion is done using SoX with my DSD patches. The source code is available at https://github.com/mansr/sox

There's an old thread about it here: http://www.computeraudiophile.com/f11-software/direct-stream-digital-encoding-sox-25556/

Link to comment
Share on other sites

TY!

The conversion is done using SoX with my DSD patches. The source code is available at https://github.com/mansr/sox

There's an old thread about it here: http://www.computeraudiophile.com/f11-software/direct-stream-digital-encoding-sox-25556/

Forrest:

Win10 i9 9900KS/GTX1060 HQPlayer4>Win10 NAA

DSD>Pavel's DSC2.6>Bent Audio TAP>

Parasound JC1>"Naked" Quad ESL63/Tannoy PS350B subs<100Hz

Link to comment
Share on other sites

This sounds like your CPU isn't quite keeping up so you're getting buffer underruns of varying duration.

 

 

 

Strange. None of those settings ought to make a difference to the CPU load. If you feel like experimenting, it would be interesting to see if you can reproduce the problem with just one of them set to the "bad" value.

 

 

 

Transitions are are tricky with DSD. Most likely, some buffer is being padded with zeros rather than DSD silence. If the DoP mode works better, I suggest using that. There is absolutely no difference to the sound while it's playing.

 

 

 

I'm not familiar with that CPU, but it doesn't seem to be exceptionally fast. It does have the AVX instruction set though, but I don't think Daphile is built to make use of this.

 

Hi mansr, I just wanted to let you know that everything works fine thanks to the latest Daphile beta I got, together with increasing my ALSA buffering value (from 1ms to 160ms), as per Kimmo's suggestions! That's fine upsampling up to DSD128.

Once I tried to upsample to DSD256, stuttering and dropouts occurred and I am guessing my cpu is not powerful enough to handle it. Doesn't really matter though, as I am in pure audio nirvana!!!

Thank you sooooo much for all your replies and feedback!! I really appreciate it!!! Be well my friend!!

 

 

Sent from my iPhone using Computer Audiophile mobile app

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share



×
×
  • Create New...