What sample rate?

[Hibster_2000]

My DAC is capable of 24 bit 192KHz.
I believe that my ears are capable of a lot less, so is there any benefit from running more than 44.1kHz sample rate?
If so, why?

[montesquieu]

In my view sample rate is (for the most part) irrelevant unless we are talking lower sample rates than redbook eg MP3 with shedloads of compression. The most important feature in any digital replay is the recording and mastering - a good quality redbook 44.1k file will sound better than a poorly recorded and mastered hi-res file regardless of the kit that it's on.

The next most important factor in digital replay is the output stage. The vast majority of CDPs and DACs use op amps of variable quality, that can make or break a component. Some of course are very good and with the right attention paid to power supplies, screening, connectors etc can sound fantastic. Others have valve output - I like to think about DACs as a preamp without a volume control - so the questions to ask are what's the quality of what's receiving the raw signal from the DAC chip at the end of the digital to analogue conversion process, adding gain, and outputting it to the connectors on the back?; what's the output impedance? essentially, is there anything constraining its performance as a preamp amplifying and outputting the converted signal to the next component in the chain?

I happen to prefer non-oversampling redbook-focused R2R chips such as the AD1865 in my own Audio Note based DAC (another example would be the TDA1541), but other than 'features' (such as the ability to play lots of different resolution files, or to manipulate the chip output in some form using digital signal processing) I think the chip is actually the least important part of the thing, certainly less important than the receiver chip/master clock/USB conversion interface and other related circuitry in bringing in the digital signal ready for analogue conversion with minimal error.

Others I'm sure will disagree but I would say focus on getting redbook right, high-res I'm sure is fun to play with but the vast majority of music out there is 41.1k and it's surely that that requires focus.

[Macca]

There's no benefit from running more than 44.1kHz sample rate.

[struth]

I try to use 24/96. I rip all my vinyl to this as well, but redbook cds are just ripped as normal although upsampled.
After saying this, when on headphones I mostly use HD bluetooth at 16/48.
The 24/96 is better but not by a lot.

[dave2010]

There's no benefit from running more than 44.1kHz sample rate.That's a bit definite.

Theoretically there's not a lot of point, but theory doesn't always work. Some - such as Grant - claim a slight improvement with more quantisation levels and a higher sampling rate. This is perhaps not because the theory (based on the sampling theorem) is incorrect, but because design factors in the recording chain and in the replay equipment do lead to an audible improvement.

A few people do genuinely seem to have hearing which can hear up to and above 20 kHz, so for them there might be a benefit in a higher sampling rate. Such people are fairly rare though, particularly among older people. I gave up worrying about TV whistle, and noise from fluorescent lights some years ago - though when I was youger these were a real pain.

[Hibster_2000]

Thanks for the replies.
I still have relatively good hearing, as small high pitch whines from led or fluorescent bulbs etc. annoy me.
I have been trying to get noise free 24 bit 192KHz output from my music PC to my SEG, but it just seems troublesome.
I have dropped the sample rate to 24/96 and this seems pretty good, and does not annoy all of the dogs in the neighbourhood so much.

[struth]

Could try one of the ifi usb solutions. They plug in in line.
The ifi silencer is 49£

[Macca]

This is perhaps not because the theory (based on the sampling theorem) is incorrect, but because design factors in the recording chain and in the replay equipment do lead to an audible improvement.

.

I'd be interested in what design factors you are taking about?

Suspect if Grant tried to distinguish his hi rez copies from his red book standard ones without knowing which is which he would...struggle. No-one in he world has ever managed to do that. And that's leaving aside the fact that vinyl does not come close to the FR or DR of red book let alone anything higher.

Always worth bearing in mind that if the recording equipment could not capture frequencies above 22Khz then there will be nothing there to reproduce even if it could be heard. And also that the loudspeakers used for playback would need to be capable of reproducing frequencies above 22Khz, and at sufficient levels to be audible.

Lots of cons in audio but 'hi res' replay has to be the biggest and most successful of them by a margin.

[struth]

in your opinion, thats fair enough. i have mine, others have theirs....such is life:)

[Hibster_2000]

Could try one of the ifi usb solutions. They plug in in line.
The ifi silencer is 49£

Thanks, but I am trying to avoid USB for audio use (rightly or wrongly).

[struth]

they do a coax one too but its expensive;)

[Sherwood]

Higher bit rates are not primarily about extending the playback frequencies. For me, the benefit of upsampling relates to how the digital signal is dealt with in the analogue processing. Try doing a google search on reconstruction filters.

I am using Audivarna to upsample my flac files and the improvement in sq is considerable. Everything just sounds smoother and more analogue.

[montesquieu]

Higher bit rates are not primarily about extending the playback frequencies. For me, the benefit of upsampling relates to how the digital signal is dealt with in the analogue processing. Try doing a google search on reconstruction filters.

I am using Audivarna to upsample my flac files and the improvement in sq is considerable. Everything just sounds smoother and more analogue.


It may sound smoother and more analogue but since no additional information is available all you are doing is manipulation of the data in the digital domain. Great that it sounds better (I'm all for that) but it has nothing whatsoever to do with a file being redbook or hi-res to begin with.

[Patrick Dixon]

I'd be interested in what design factors you are taking about?


I imagine that he's talking about the effects of anti-alias filtering before sampling, and of reconstruction filtering after conversion back to analogue.

Deficiencies in either have the potential to alias (unwanted) high frequencies onto low frequencies so they can produce in-audio-band frequencies from out-of-audio-band frequencies.

If you sample at 44.1KHz any energy above 22.05KHz is reflected back about 44.1KHz, so (eg) a 30KHz signal would appear at 14.10KHz (with opposite phase). You can see this effect in old western films (shot at 25fps) where a wagon wheel can start off rotating in the right direction, and then appear to rotate backwards when the wagon reaches a certain speed.

Good design should avoid this, but the higher the sampling rate the more room the designer has to remove the unwanted signal without harming the ones you do want to retain.

[montesquieu]

I imagine that he's talking about the effects of anti-alias filtering before sampling, and of reconstruction filtering after conversion back to analogue.

Deficiencies in either have the potential to alias (unwanted) high frequencies onto low frequencies so they can produce in-audio-band frequencies from out-of-audio-band frequencies.

If you sample at 44.1KHz any energy above 22.05KHz is reflected back about 44.1KHz, so (eg) a 30KHz signal would appear at 14.10KHz (with opposite phase). You can see this effect in old western films (shot at 25fps) where a wagon wheel can start off rotating in the right direction, and then appear to rotate backwards when the wagon reaches a certain speed.

Good design should avoid this, but the higher the sampling rate the more room the designer has to remove the unwanted signal without harming the ones you do want to retain.

Debates have long raged about upsampling and filtering - one of those ideological battles in my opinion - but the question was about playback of redbook vs higher resolution files, ie files recorded and mastered at higher bit rates. This has nothing do with upsampling which is after all just a method of manipulating data that's fed in to the system from a disk, file or stream, whatever resolution it starts at.

[Sherwood]

I imagine that he's talking about the effects of anti-alias filtering before sampling, and of reconstruction filtering after conversion back to analogue.

Deficiencies in either have the potential to alias (unwanted) high frequencies onto low frequencies so they can produce in-audio-band frequencies from out-of-audio-band frequencies.

If you sample at 44.1KHz any energy above 22.05KHz is reflected back about 44.1KHz, so (eg) a 30KHz signal would appear at 14.10KHz (with opposite phase). You can see this effect in old western films (shot at 25fps) where a wagon wheel can start off rotating in the right direction, and then appear to rotate backwards when the wagon reaches a certain speed.

Good design should avoid this, but the higher the sampling rate the more room the designer has to remove the unwanted signal without harming the ones you do want to retain.

Exactly, though I'm not sure about the wagon wheel analogy. I am not interested in ideological debates only sq. I have only recently become a convert to digital and I am sure that my distaste had a lot to do with poor digital processing. Using Audivarna on a reasonably powerful pc allows me to upsample redbook and even high res files. The improvement in sq is not small.

[dave2010]

Always worth bearing in mind that if the recording equipment could not capture frequencies above 22Khz then there will be nothing there to reproduce even if it could be heard. And also that the loudspeakers used for playback would need to be capable of reproducing frequencies above 22Khz, and at sufficient levels to be audible.

Lots of cons in audio but 'hi res' replay has to be the biggest and most successful of them by a margin.Yet another problem would be if the recording equipment could capture frequencies above 22 kHz, then depending on how the data was processed those sounds could be mapped down into the audible range and would appear as noise - which may, or may not, be annoying. If that is going to be a problem (bats flying round the recording studio - though to my slight surprise I actually heard a bat a year or so back - I didn't think my hearing went that high ..) then the high frequencies should be filtered off either with an analogue filter, or a digital one before the signal is converted to the digital form for distribution.

[Patrick Dixon]

Debates have long raged about upsampling and filtering - one of those ideological battles in my opinion - but the question was about playback of redbook vs higher resolution files, ie files recorded and mastered at higher bit rates. This has nothing do with upsampling which is after all just a method of manipulating data that's fed in to the system from a disk, file or stream, whatever resolution it starts at.

I don't think I mentioned upsampling. My post has nothing to do with upsampling either!

[Patrick Dixon]

Exactly, though I'm not sure about the wagon wheel analogy.
It's not an analogy, it's an example of aliasing.

[Sherwood]

It's not an analogy, it's an example of aliasing.

If we are being pedantic it is both as you are making comparisons between audio and video (analogy def: a comparison (https://dictionary.cambridge.org/dictionary/english/comparison) between things that have similar (https://dictionary.cambridge.org/dictionary/english/similar) features (https://dictionary.cambridge.org/dictionary/english/feature), often used to help (https://dictionary.cambridge.org/dictionary/english/help) explain (https://dictionary.cambridge.org/dictionary/english/explain) a principle (https://dictionary.cambridge.org/dictionary/english/principle) or idea (https://dictionary.cambridge.org/dictionary/english/idea))

[Sherwood]

i am puzzled by those who write off the benefits of high definition files and upsampling. Regarding the latter, consider a vinyl pressing. It is not possible to add more data to the grooves once it is pressed. However, one can extract more of the data in the grooves by improving the replay chain. A better RIAA device, superior cartridge, more rigid tonearm, better power supply, more highly engineered bearings etc. The same is true of redbook files. I find that a system that can perform a good upsampling of redbook files typically sounds "better" (to me at least) than one that does not. I find a similar improvement in files that have been recorded in hd. If others don't perceive the same improvements that is fine.

[montesquieu]

i am puzzled by those who write off the benefits of high definition files and upsampling. Regarding the latter, consider a vinyl pressing. It is not possible to add more data to the grooves once it is pressed. However, one can extract more of the data in the grooves by improving the replay chain. A better RIAA device, superior cartridge, more rigid tonearm, better power supply, more highly engineered bearings etc. The same is true of redbook files. I find that a system that can perform a good upsampling of redbook files typically sounds "better" (to me at least) than one that does not. I find a similar improvement in files that have been recorded in hd. If others don't perceive the same improvements that is fine.

That's my point though. A file recorded, mastered and distributed at hi-res (where this is actually some additional resolution, in theory anyway) is or at least ought to be a completely different animal to a lower res file upsampled. Whatever your kit is doing to process the signal may be the important thing here in why you like it, not the initial (or indeed the upsampled) resolution. Great if you like it but there's no proof whatsoever that a file being hi-res is responsible for why you like it.

[Macca]

i am puzzled by those who write off the benefits of high definition files and upsampling. Regarding the latter, consider a vinyl pressing. It is not possible to add more data to the grooves once it is pressed. However, one can extract more of the data in the grooves by improving the replay chain. A better RIAA device, superior cartridge, more rigid tonearm, better power supply, more highly engineered bearings etc. The same is true of redbook files. I find that a system that can perform a good upsampling of redbook files typically sounds "better" (to me at least) than one that does not. I find a similar improvement in files that have been recorded in hd. If others don't perceive the same improvements that is fine.

There's no data in the groove of a vinyl record though, it's just a groove. Man. :)

I was an early adopter of SACD. Initially I thought it was the second coming. With the hybrid discs I was puzzled that the CD layer and SACD layer sounded no different. Then I found out.

Also had one of the upsampling players. Didn't like it, sounded a bit blousey, thought the upsampling was to blame. With CD you could not switch off the upsampling. But it could be used as a stand alone DAC too so I hooked up a transport and could switch between 16/44.1 and 24/384 at will with the remote. Not a scrap of difference.

Not to say that up sampling will not make a difference sometimes. Depends on all sorts of things. Lots of variables involved.

[Edward]

I find this article (https://people.xiph.org/~xiphmont/demo/neil-young.html)helps keep me on track regarding high resolution audio.

[struth]

It doesn't always help, but it can.
Like most things that are complicated,... It's complicated.

[montesquieu]

I find this article (https://people.xiph.org/~xiphmont/demo/neil-young.html)helps keep me on track regarding high resolution audio.

Thanks Edward I'd read that before but I'll bookmark it this time.

[Patrick Dixon]

If we are being pedantic it is both as you are making comparisons between audio and video (analogy def: a comparison (https://dictionary.cambridge.org/dictionary/english/comparison) between things that have similar (https://dictionary.cambridge.org/dictionary/english/similar) features (https://dictionary.cambridge.org/dictionary/english/feature), often used to help (https://dictionary.cambridge.org/dictionary/english/help) explain (https://dictionary.cambridge.org/dictionary/english/explain) a principle (https://dictionary.cambridge.org/dictionary/english/principle) or idea (https://dictionary.cambridge.org/dictionary/english/idea))
Sorry but you are incorrect; it's an example of sampling something with a frequency lower than half the highest frequency that you are sampling, which produces aliasing. It's not an analogy because I am not making a comparison between things that have similar features, I am giving an example of the phenomenon (aliasing).

Anyway, now you fully understand aliasing, it's job done AFAIAC.

[Sherwood]

Sorry but you are incorrect; it's an example of sampling something with a frequency lower than half the highest frequency that you are sampling, which produces aliasing. It's not an analogy because I am not making a comparison between things that have similar features, I am giving an example of the phenomenon (aliasing).

Anyway, now you fully understand aliasing, it's job done AFAIAC.

You have the definition: if you can't grasp it then that's fine.

[mikmas]

Sorry but you are incorrect; it's an example of sampling something with a frequency lower than half the highest frequency that you are sampling, which produces aliasing. It's not an analogy because I am not making a comparison between things that have similar features, I am giving an example of the phenomenon (aliasing).
.

Actually Geoff was absolutely correct - the comparison you made was an analogy (by its very definition...).

The phenomenon of the 'wagon wheel' rotation and film projection (not video) has very little to do with digital 'aliasing' other than both being related to frequency differences that produce an undesirable effect; thus wholly analogous.

The two effects are not even remotely connected either technologically or in their underlying physics.

[Gazjam]

The important bit (in my opinion...having dabbled with digital a bit) is the bit depth, not the sample rate.
A 24bit 44Khz version of a song has always sounded noticeably better than even a 352Khz 16 bit up-sample of the same track.

Bit depth is what matters...

[Gazjam]

There's no benefit from running more than 44.1kHz sample rate.

Only the Sith deal in absolutes...

[Macca]

Only the Sith deal in absolutes...

Who says I'm not a Sith? :D

I don't get your thing about bit depth. All bit depth does is determine the maximum dynamic range possible. 16 bits gives more dynamic range than you need for 99.99% of recordings and even then it is probably higher than the amplifier is capable of anyway - what is the weakest link with regard to dynamic range in any system? It isn't going to be the digital source even if it is a supermarket brand DVD player.

All very well to talk about the effects of aliasing, bit depth, pre and post ringing, jitter etc, but can you hear them when music is playing? And the answer is no. If these things were audible vinyl would sound terrible since its sins are of far worse magnitude in these respects.

[Edward]

Archimago did an interesting experiment to demonstrate that even with high jitter and drift values you can't hear the differences. He created versions of a track that had jitter and drift values far in excess of even poorly designed DACs.

This is his write up (http://archimago.blogspot.com/2019/08/musings-demo-why-bits-are-bits-lets-not.html) and the relevant tracks can be obtained here (http://redirect.viglink.com/?format=go&jsonp=vglnk_157269837242516&key=a7cf329d5bd13115736c5131a3c772ff&libId=k2hjrnf50100rcxm000DLbg54lz5e&loc=http%3A%2F%2Farchimago.blogspot.com%2F2019%2F0 8%2Fmusings-demo-why-bits-are-bits-lets-not.html&v=1&opt=true&out=https%3A%2F%2Fwww.amazon.ca%2Fclouddrive%2Fsha re%2FE74h7t6D7pzveiQ1ohLWq8s83Y437QnVyzZIig1Y8DA&ref=https%3A%2F%2Fwww.google.com%2F&title=Archimago%27s%20Musings%3A%20MUSINGS%20%2F%2 0DEMO%3A%20Why%20%22Bits%20Are%20Bits%22.%20Let%27 s%20not%20add%20unnecessary%20fear%2C%20uncertaint y%2C%20and%20doubt.&txt=Archimago%20-%20Drift%20and%20Jitter%20Demo.zip).

Anyway I certainly can't hear any differences - even after several plays. I was keen to hear differences so as to confirm my positive bias I have towards my set up. It is not to be.

[struth]

All depends if you suffer from Al Gore isms:eyebrows:

[Edward]

The important bit (in my opinion...having dabbled with digital a bit) is the important part is the bit depth, not the sample rate.
A 24bit 44Khz version of a song has always sounded noticeably better than even a 352Khz 16 bit up-sample of the same track.

Bit depth is what matters...

Definitely bit depth matters. However if you upsampled a 16 bit track to 352Khz (or whatever frequency) whilst staying with 16 bit wordlength the resultant track will sound worse due to rounding. Any digital manipulation needs to be done at a high enough bit depth (typically DSP these days do things at 64 bit floating) and then dithered down to the end result (16, 24 etc). Your 24 bit 44.1khz track will sound better then the 16 bit upsampled track.

Or did I misunderstand your post? :scratch:

Of course the upsampling engine you used may be at fault.

[struth]

been using dsd in main system as audirvana does it and pc over there has enough memory etc so there is no problems

[dave2010]

Definitely bit depth matters. However if you upsampled a 16 bit track to 352Khz (or whatever frequency) whilst staying with 16 bit wordlength the resultant track will sound worse due to rounding. Any digital manipulation needs to be done at a high enough bit depth (typically DSP these days do things at 64 bit floating) and then dithered down to the end result (16, 24 etc). Your 24 bit 44.1khz track will sound better then the 16 bit upsampled track.
I think your comments are theoretically correct - but some people might find that practically they're not - for them at least. I can't actually comment sufficiently well on audio, but I have done work with video and image files. Much of the theory is the same, but sometimes it is definitely possible to get a much better visual effect with some digital processing, and perhaps also upsampling. Compared with the original images the modified ones must have been less good, because of the processing, but most people would prefer the edited ones.

There are limits to what can be achieved, however, sometimes the improvements are only slight if there's insufficient detail in the originals.

[Sherwood]

I think your comments are theoretically correct - but some people might find that practically they're not - for them at least. I can't actually comment sufficiently well on audio, but I have done work with video and image files. Much of the theory is the same, but sometimes it is definitely possible to get a much better visual effect with some digital processing, and perhaps also upsampling. Compared with the original images the modified ones must have been less good, because of the processing, but most people would prefer the edited ones.

There are limits to what can be achieved, however, sometimes the improvements are only slight if there's insufficient detail in the originals.

That's the key point. When the vinyl record was developed I am sure that few people could imagine how much information could be recorded in the grooves. At the time, the technology to retrieve the Nth degree of detail from a groove was not fully developed. Over the years, there has been a progressive improvement in technology and engineering so that albums recorded and pressed 50 years ago (or longer) can sound fantastic. Yes, there were very good reproduction systems around by the 60's, but they were the exception rather than the norm. I believe the same applies to digital. When the CD was launched I contend that the same was true: that there was an imbalance in the audio chain and that encoding analogue to digital was more successful than the reverse process. I could barely listen to cd when it was first launched. I was particularly sensitive to the high frequency playback and what I now perceive to be "time smearing". Again, there were a few decent cd players around, but most were way out of my budget. Over time, these problems seem to have been addressed and I am now happy to have all but switched over to digital. However, it has taken a long time to make good quality digital reproduction truly affordable and for me the biggest improvement has come with the better processing of digital to analogue through upsampling.

[Gazjam]

Definitely bit depth matters. However if you upsampled a 16 bit track to 352Khz (or whatever frequency) whilst staying with 16 bit wordlength the resultant track will sound worse due to rounding. Any digital manipulation needs to be done at a high enough bit depth (typically DSP these days do things at 64 bit floating) and then dithered down to the end result (16, 24 etc). Your 24 bit 44.1khz track will sound better then the 16 bit upsampled track.

Or did I misunderstand your post? :scratch:

Of course the upsampling engine you used may be at fault.

You just agreed with me without realising it.
Cool.

[Edward]

I think your comments are theoretically correct - but some people might find that practically they're not - for them at least. I can't actually comment sufficiently well on audio, but I have done work with video and image files. Much of the theory is the same, but sometimes it is definitely possible to get a much better visual effect with some digital processing, and perhaps also upsampling. Compared with the original images the modified ones must have been less good, because of the processing, but most people would prefer the edited ones.

There are limits to what can be achieved, however, sometimes the improvements are only slight if there's insufficient detail in the originals.

Hi Dave, I agree about your main point about possible improvements via upsampling (I to do image processing) but my narrow point was about doing upsampling (or indeed any processing) whilst staying in the 16 bit domain.

[Edward]

You just agreed with me without realising it.
Cool.


Of course I realised I agree with you. :)

My point is why would you stay in the 16 bit domain when upsampling if you knew the sq will be rubbish.

[struth]

think you often see that with qobuz, as many you would maybe think would be done in hi rez are kept at 16/44. also you get others at odd levels so obviously they are adj level to suit the sound quality

[dave2010]

There's no data in the groove of a vinyl record though, it's just a groove. Man. :)

I was an early adopter of SACD. Initially I thought it was the second coming. With the hybrid discs I was puzzled that the CD layer and SACD layer sounded no different. Then I found out.

Also had one of the upsampling players. Didn't like it, sounded a bit blousey, thought the upsampling was to blame. With CD you could not switch off the upsampling. But it could be used as a stand alone DAC too so I hooked up a transport and could switch between 16/44.1 and 24/384 at will with the remote. Not a scrap of difference.

Not to say that up sampling will not make a difference sometimes. Depends on all sorts of things. Lots of variables involved.
This probably won't be your cup of tea, but over the last year I've got into playing different recorders (the blowy things, the 'orrible squeaky things which some 3-8 year olds like to annoy you with right down yer lug'oles ....).

Seems that "real" players can tell the difference between models - and yes - now that I've got more than ten different models (though only one bass recorder - which is just passable). From a player's perspective I can say that there are differences. In order to decide which of some of the models to buy I looked at the Thomann web site - such as this - https://www.thomann.de/gb/yamaha_yrt304b.htm
You should notice that there are audio samples, which are supposed to be representative of the model.
Compare that one, for example with this one https://www.thomann.de/gb/aulos_211a_robin_tenor.htm or even this one https://www.thomann.de/gb/thomann_trt31b_tenor_blockfloete.htm

It is possible that you won't hear much difference, particularly if you're not a player. I actually do have one of the Aulos models - it's OK, and one of the Thomann models - which is really not great. I also played a Yamaha which I borrowed - and perhaps it did sound the best on the Thomann web site - though like Thomann's own instrument, it has fussy keys which don't work too well.

Another comparison site is this one - https://youtu.be/DEJA49rZyCs - you won't find out which is which until much later, if you ever do.

A three way comparison is here - https://youtu.be/tnTimwvHmwM

Sometimes the audible differences seem very small - though that may also depend on the state of one's ears. Sometimes putting olive oil in the ears can make a significant difference to how things sound - sometimes good - and sometimes not good at all. I am prone to ear wax, so sometimes I get a big problem, though often now I just leave things alone and problems do quite often just go away. I do, however, think that extended listening on headphones can make wax and hearing problems much worse.

If you can't hear all the resolution which others claim you should, don't worry about it, unless it bothers you.

[Gazjam]

Of course I realised I agree with you. :)

My point is why would you stay in the 16 bit domain when upsampling if you knew the sq will be rubbish.

I dont. :)

Dabbled with upsampling, HQ Player and the like, but nowadays prefer playback in native format.
My Dac upsamples to DSD anyway, so upsampling is moot in my case.

[Hibster_2000]

So it would appear that sample rate questions are the new cable thread?
I have dropped to 24/96 and it sounds fine.

[NRG]

Hi Dave, I agree about your main point about possible improvements via upsampling (I to do image processing) but my narrow point was about doing upsampling (or indeed any processing) whilst staying in the 16 bit domain.

Yeah, upsampling is a frequency change IE inserting samples not a bit depth change. A greater bit depth helps when manipulating digital files like when changing peak levels after which you dither to the required bit depth.

Patrick was shat upon earlier and unfairly, he knows his stuff, the analogy was a good one given a sampling rate (24fps) of a none BANDWIDTH LIMITED input. The resulting aliasing being seen as a wheel rotating backwards.

[dave2010]

Yeah, upsampling is a frequency change IE inserting samples not a bit depth change. A greater bit depth helps when manipulating digital files like when changing peak levels after which you dither to the required bit depth.
If upsampling is used, with interpolation then arbitrary data values may be obtained. If using floating point arithmetic these can be mapped to any bit depth by truncation, though the preferred method would be to use dither.

Whether this could realistically lead to any perceived sound improvement I couldn't say.

Once one starts messing with filtering etc. all sorts of perceived "improvements" can be noticed. In image processing I have been amazed at some of the effects - one being haze removal. The point of the image processing analogy is to show that "improvements" are possible. Such improvements may not actually correspond to reality - it may actually have been hazy in the example mentioned - but then one can open up all sorts of discussion about what "reality" is anyway.

These perceptual changes emerge after either linear (most common?) or non-linear filtering. Doing manipulation of audio signals using digital processing can use filtering explicitly, but will usually change the data very slightly (unless true lossless procesing is used), and the changes may be perceived as improvements, or not. Such changes, which strictly should be considered as noise, will also interact with the analogue circuitry used for the playback.

Someone up thread suggests "too many variables"!

[Macca]

Someone up thread suggests "too many variables"!

Yes, too many variables to be able to deduce purely from listening as to what is causing the outputted signal to be different. Or even if it is actually different. That doesn't seem to stop people from jumping to conclusions or coming out with some pretty outrageous theories as to why. But it's all good fun.

[struth]

like most things for the end user, what sounds best is best . folk can overthink things and spoil what should be pleasure of discovery;)

[dave2010]

Yes, too many variables to be able to deduce purely from listening as to what is causing the outputted signal to be different. Or even if it is actually different. That doesn't seem to stop people from jumping to conclusions or coming out with some pretty outrageous theories as to why. But it's all good fun.Indeed. A few years ago I looked at some waveforms from clarinets, and simulated clarinets. The odd thing was that some of the simulated clarinets had really odd waveforms, yet sounded much closer to so some of the real clarinet samples, while some of the real clarinet samples sounded really quite different from some of the others. I was a bit surprised, but that's the way things are.

Despite being aware of some obvious deficiencies in my hearing compared with many years ago, I continue to be amazed on occasions at how much better and realistic (!!) real live sound by live performers in an appropriate environment is, compared with even reasonably good "hi-fi". Of course, sometimes it can be worse, but not often. It's usually much better by several orders of magnitude That applies to almost all forms of music, or aural experiences.