tuga
Legal Alien
As far as I understand the goal of resampling to very high rates is to improve reconstruction at Nyquist, NOT to recover data which isn't there.
Upsampling
- Thread starter Alex S
- Start date
As far as I understand the goal of resampling to very high rates is to improve reconstruction at Nyquist, NOT to recover data which isn't there.
tuga
Legal Alien
Indeed microphones & speakers+rooms are so imperfect, why do we need electronics with less than 1% THD is something I never understood, or anything better than mp3 or even shellac? Surely anything above that isn't audible...
Also, my theory being wrong it seems daft to oversample 16/44.1...
Jim Audiomisc
pfm Member
Yes. The point being that this may be a more practical way to do things like the filtering. It also can ease dealing with things like ADC nonlinearities, etc. Hence, for example, the first Philips chipset that oversampled x4. Helped the filtering and eased the effects of the limited DAC hardware in terms of processed sample sizes, monotonicity, etc.
So overampling can be very useful.
Jim Audiomisc
pfm Member
The point is that theory and reality aren't usually the same.
Of course, in theory they are, but reality may not have read that book.... 8-]
nostromo
pfm Member
I'm finding it almost impossible to follow what's been said here, although I'm interested.
I need a simple answer, should I oversample? I'm using Audirvana with an NOS DAC (Holo Spring DAC). If yes to what level?
I don't think I can hear a difference with previous tinkering, but perhaps my guests can so I just want to set it up 'right' and forget about it.
I need a simple answer, should I oversample? I'm using Audirvana with an NOS DAC (Holo Spring DAC). If yes to what level?
I don't think I can hear a difference with previous tinkering, but perhaps my guests can so I just want to set it up 'right' and forget about it.
Jim Audiomisc
pfm Member
If you don't hear a difference but want to know, then simply ask your 'guests' if they do. And if so, which they prefer, or if they think it is too small a change to care about. Ideally, not telling them which version is oversampled and which isn't.
tuga
Legal Alien
You should oversample/filter, unless you prefer the potentially euphonic effects of not using a reconstruction filter (ultimately it's your call).
The Spring performs best when fed DSD256, or you could try 20-bit 705.6/768kHz PCM.
Alex S
carbon based lifeform
I think it’s worth trying two settings: a factor of two and to the max. Not much faffing required, and see if your guests hear any difference.
There are long threads elsewhere proclaiming the benefit for the Holo but don’t tell your guests that.
naka
pfm Member
I am guessing that your Audial S4 (TDA 1541) sounds sublime… without any upsampling.
My Musical Fidelity AMS CD player / Dac, automatically upsamples everything to 24/192 and it is great.
But also is the Opera Consonance DAC16 (TDA1543) using the MF as CD transport.
My experiences over the years led me to some disinterest about upsampling… until now, maybe one day
gustav_errata
pfm Member
That may have been a problem in the past but these days most digital audio workstations and certainly all professional mastering engineers have easy access to true peak (dBTP) meters.
gustav_errata
pfm Member
No matter what genre of music you listen to, every step of the recording and mixing process was subjected to qualitative decisions based on what the engineers felt best capture the sound in a way that they like: the studio/hall, the microphones, the preamplifiers, the mic compressors, the EQ, the mixing board, and even, yes, the conversion ( for the longest time many people felt that Appogee made the best sounding, "must musical" ADC with its soft clipping facility). Add to that whatever further processing is done at the mixing stage. Only at the mastering stage do they aim for the cleanest processing and conversion so as not to impart flavour (in the subtle imperfect hardware "mojo" sense).
So, if one's objective is to reproduce the performance accurately, it's a fool's errand. If one's objective is to reproduce the recorded signal as accurately as possible, then upsampling to remove even the remotest concerns about the DAC filtering or about digital aliasing in any pre-DAC DSP, is surely a good idea. If one's objective is produce an imagined music performance in the studio/venue of their listening room, then try to find whatever combination of DSP, sampling, and filter settings floats your boat.
Sloop John B
And any old music will do…
This ^^
Rarely heard night and day differences with DACs. Not to many tried, Naim DAC, Hugo TT, Hugo TT2/mscaler and Holo Audio Spring.
They are so easy to slot in and out I think we are all tempted to tweak, whereas speakers are big heavy beasts which are much more difficult to move on.
.sjb
darrenyeats
pfm Member
Some resampling algorithms are just more precise/accurate (lower distortion products) than others. See https://src.infinitewave.ca/ with 1kHz measurement.
E.g. Signalist 2.9.1, Weiss Saracon and others seem to be impeccable, some others are very good - some others not.
Given that even computer software can vary in quality, it's plausible computer upsampling - if done optimally - could simply be more precise/accurate than some DAC implementations.
Whether this aspect can be audible is another matter but it's measurable at least.
E.g. Signalist 2.9.1, Weiss Saracon and others seem to be impeccable, some others are very good - some others not.
Given that even computer software can vary in quality, it's plausible computer upsampling - if done optimally - could simply be more precise/accurate than some DAC implementations.
Whether this aspect can be audible is another matter but it's measurable at least.
darrenyeats
pfm Member
Implementation is key. And I'm sure some DACs have excellent in-built upsampling.
Out of curiousity which one do you have?
I am also looking for a NOS DAC and appreciate any input.
Thanks
MarkW
Full Speed & Pagan
Nostromo
My take after many years of futzing about, reading and listening:
Yes, theory definitely says you should upsample. Correctly done (and there's no telling if a given filter is good enough) it will improve the reconstruction of the original signal over not upsampling the same file. The brutal truth is that the samples contain ALL the information that was in the original signal WITHIN THE PASSBAND of the adc filter used to digitize the signal but without passing those samples through a reconstruction filter recovering that information is not possible because of the artifacts of the analogue to digital filter process and the noise that is an outcome of the process.
I have heard some lovely sounding NOS dacs (and some bloody awful oversampling ones) but they are basically doing it wrong (according to theory).
For what its worth, I am up to my ears in the boggy upsampling camp (see the Dac200 thread for the gory details) and I find significantly more music in pushing the process to the nth degree. Just regular hardware upsampling (Limited resources, even with a Chord Dave) or perfunctory software (pretty much everything less than HQPLayer 4 or PGGB) is pretty much indistringuishable from not upsampling.
A Holo Sping or May (not Denafrips, many are convinced there is some processing going on) if a well engineered product - feeding it good upsampled bits should produce better results than basic bits.
My take after many years of futzing about, reading and listening:
Yes, theory definitely says you should upsample. Correctly done (and there's no telling if a given filter is good enough) it will improve the reconstruction of the original signal over not upsampling the same file. The brutal truth is that the samples contain ALL the information that was in the original signal WITHIN THE PASSBAND of the adc filter used to digitize the signal but without passing those samples through a reconstruction filter recovering that information is not possible because of the artifacts of the analogue to digital filter process and the noise that is an outcome of the process.
I have heard some lovely sounding NOS dacs (and some bloody awful oversampling ones) but they are basically doing it wrong (according to theory).
For what its worth, I am up to my ears in the boggy upsampling camp (see the Dac200 thread for the gory details) and I find significantly more music in pushing the process to the nth degree. Just regular hardware upsampling (Limited resources, even with a Chord Dave) or perfunctory software (pretty much everything less than HQPLayer 4 or PGGB) is pretty much indistringuishable from not upsampling.
A Holo Sping or May (not Denafrips, many are convinced there is some processing going on) if a well engineered product - feeding it good upsampled bits should produce better results than basic bits.
tuga
Legal Alien
To the max (32/1.536PCM or DSD1024) will require an extremely powerful computer.
tuga
Legal Alien
That may be due to poor implementation, poor algorithms or bcause the processing in the DAC is generating rubbish.
Measurements from the DAC in question, Holo Spring 3:
tuga
Legal Alien
These are from a TEAC UD-501 set to:
44.1/16 NOS,
44.1/16 SHARP,
44.1/16 SLOW,
44.1/16 upsampled to 384/32 into NOS,
44.1/16 upsampled to 384/32 into NOS,
44.1/16 upconverted to DSD128 into NOS
(unfortunately the page formatting has gone mad, but it's easy to get the picture)
audiophilestyle.com
44.1/16 NOS,
44.1/16 SHARP,
44.1/16 SLOW,
44.1/16 upsampled to 384/32 into NOS,
44.1/16 upsampled to 384/32 into NOS,
44.1/16 upconverted to DSD128 into NOS
(unfortunately the page formatting has gone mad, but it's easy to get the picture)
Playback formats on a TEAC UD-501 (TI PCM1795)
This is interesting DAC with a reasonable price tag, looks nice and high quality components. Since it also offers bypass of the digital interpolation filters with direct 32-bit 352.8/384 PCM input as well as DSD128 support with good set of analog filters (conversion mappings), I wanted to study w...
audiophilestyle.com