[RRod]

  My DAC is an oversampling DAC. It actually doesn't even give you an option: 24/96 or 24/192, no "none". I read a while back that it will take a 24/96 or 192 file, downscale it to 16/44, and then upscale it again. In addition to that, it doesn't support DSD, and I'll want to try out DSD and hi-res PCM in the future. Should I avoid or favor "oversampling"? What's the pros and cons?

 
The point of oversampling is to make life easier on analog components by moving work to the digital domain. At the end of the day the point is still to recreate the original signal post-band-limiting.

 

[Speedskater]

   
Level matching  is only one of the typical problems with audiophile sighted casual evaluations of CD versus LP or tube versus solid state or amplifier versus amplifier.
(1) Audiophile Sighted Casual Evaluations are not reliable evidence because they are not tests. That is, they do not involve comparison to a fixed, reliable standard.

(2) Audiophile Sighted Casual Evaluations are not reliable evidence because they involve excessively long switchover times, which makes them highly susceptible to false negatives because they desensitize the listeners.

(3) Audiophile Sighted Casual Evaluations are not reliable evidence  because the do not involve proper level matching, which makes them highly susceptible to false positives because people report the level mismatches as sonic differences.

(4) Audiophile Sighted Casual Evaluations are not reliable evidence because they do not involve listening to the identical same piece of music or drama within a few milliseconds, creating false positives because people report the mismatched music as sonic differences in the equipment.

(5) Audiophile Sighted Casual Evaluations are not reliable evidence because they constantly reveal the true identity of the UUTs to the listener, creating false positives because people report their prejudices and preconceived notions as sonic properties of the equipment

Also true are:
 
Level matching  is only one of the typical problems with audiophile blind casual evaluations of CD versus LP or tube versus solid state or amplifier versus amplifier.
 
(1) Audiophile Blind Casual Evaluations are not reliable evidence because they are not tests. That is, they do not involve comparison to a fixed, reliable standard.

 

(2) Audiophile Blind Casual Evaluations are not reliable evidence because they involve excessively long switchover times, which makes them highly susceptible to false negatives because they desensitize the listeners.

 

(3) Audiophile Blind Casual Evaluations are not reliable evidence  because the do not involve proper level matching, which makes them highly susceptible to false positives because people report the level mismatches as sonic differences.

 

(4) Audiophile Blind Casual Evaluations are not reliable evidence because they do not involve listening to the identical same piece of music or drama within a few milliseconds, creating false positives because people report the mismatched music as sonic differences in the equipment.

***************************************************

Now this one might depend on just how blind the test was.

(5) Audiophile Sighted Casual Evaluations are not reliable evidence because they constantly reveal the true identity of the UUTs to the listener, creating false positives because people report their prejudices and preconceived notions as sonic properties of the equipment

 

[Pio2001]

Resampling, oversampling and ASRC (Asynchronous Sample Rate Conversion) are three different things.
 
Oversampling is done natively inside nearly all DACs. It is a necessary step for a proper digital-to-analog conversion, especially at 44100 Hz. Some exotic high-end designs are capable of converting to analog without any oversampling (the "NOS" ones). This is a mistake, coming from a complete misunderstanding of the basic science of digital PCM. They believe that oversampling is "something" on the path of the signal and that it is better to "remove" it.
It is the same thing as saying that the crossover of a 2-way speaker is something in the path of the signal and that it should be removed.
 
Resampling is changing the sample rate of the digital data before sending it to the DAC. It occurs in computers, or devices that need to mix together several tracks before sending them to a DAC. They convert everything to, say, 48 kHz, then they sum each sample to mix the tracks together, then they play the resulting 48 kHz stream. In many cases, they keep converting to 48 kHz even when they just playback a single 44.1 kHz track. It is possible to do this nearly without quality loss, but not all devices do this well.
 
Last, ASRC is a sample rate conversion whose rate can vary in realtime. It allows to mix two S/Pdif streams together, for example. Since both streams have their own clock, they slowly drift from one another. ASRC  resamples them to another independant clock. It is extremely difficult to do it without quality loss, because the speed must be adjusted by extremely small increments, and converting from, say 44099 Hz to 44100 Hz produces a lot of aliasing.
In computers, the VLC video player has built-in ASRC. As such, the speed can change audibly during playback. Turning off ASRC while playing local files is a feature that has been asked for years to the VLC team (it is only needed to playback realtime streaming, such as radio stations), but no one has volunteered yet to do the job. It was also used in the Reclock filter, that allowed to play back DVD in computers following the video clock instead of the audio clock, to avoid micro-judder in the picture. This one could adjust the speed more precisely than VLC, but in counterpart, it introduced clearly audible aliasing in the sound (the triangle of the orchestra in 2001 space odyssey made "tiwiwiwink" instead of "tinnnnnk").
ASRC must be avoided in high fidelity. It may be used in "anti-jitter" devices. These little boxes pretend to eliminate the jitter that comes from a CD Drive feeding an external DAC. This is nonsense. First, DACs eliminate all the jitter from their digital input. All that's left in the analog output comes from the DAC itself. Second, there is no reason for the drive to have more jitter than the anti-jitter device itself. Third, if the device uses ASRC (some may just use PLLs), it harms the sound quality much more than any amount of jitter could.

 

[jtaylor991]

 
Also true are:
 
Level matching  is only one of the typical problems with audiophile blind casual evaluations of CD versus LP or tube versus solid state or amplifier versus amplifier.
 
(1) Audiophile Blind Casual Evaluations are not reliable evidence because they are not tests. That is, they do not involve comparison to a fixed, reliable standard.

 

(2) Audiophile Blind Casual Evaluations are not reliable evidence because they involve excessively long switchover times, which makes them highly susceptible to false negatives because they desensitize the listeners.

 

(3) Audiophile Blind Casual Evaluations are not reliable evidence  because the do not involve proper level matching, which makes them highly susceptible to false positives because people report the level mismatches as sonic differences.

 

(4) Audiophile Blind Casual Evaluations are not reliable evidence because they do not involve listening to the identical same piece of music or drama within a few milliseconds, creating false positives because people report the mismatched music as sonic differences in the equipment.

***************************************************

Now this one might depend on just how blind the test was.

(5) Audiophile Sighted Casual Evaluations are not reliable evidence because they constantly reveal the true identity of the UUTs to the listener, creating false positives because people report their prejudices and preconceived notions as sonic properties of the equipment

 
So, blind or sighted, audiophile casual evaluations are unreliable then? What's the replacement (i.e. "non-casual")?
 
Is this a recommendation that I forget about it (comparing by ear at all), or..?
 
 

  Resampling, oversampling and ASRC (Asynchronous Sample Rate Conversion) are three different things.
 
Oversampling is done natively inside nearly all DACs. It is a necessary step for a proper digital-to-analog conversion, especially at 44100 Hz. Some exotic high-end designs are capable of converting to analog without any oversampling (the "NOS" ones). This is a mistake, coming from a complete misunderstanding of the basic science of digital PCM. They believe that oversampling is "something" on the path of the signal and that it is better to "remove" it.
It is the same thing as saying that the crossover of a 2-way speaker is something in the path of the signal and that it should be removed.
 
Resampling is changing the sample rate of the digital data before sending it to the DAC. It occurs in computers, or devices that need to mix together several tracks before sending them to a DAC. They convert everything to, say, 48 kHz, then they sum each sample to mix the tracks together, then they play the resulting 48 kHz stream. In many cases, they keep converting to 48 kHz even when they just playback a single 44.1 kHz track. It is possible to do this nearly without quality loss, but not all devices do this well.
 
Last, ASRC is a sample rate conversion whose rate can vary in realtime. It allows to mix two S/Pdif streams together, for example. Since both streams have their own clock, they slowly drift from one another. ASRC  resamples them to another independant clock. It is extremely difficult to do it without quality loss, because the speed must be adjusted by extremely small increments, and converting from, say 44099 Hz to 44100 Hz produces a lot of aliasing.
In computers, the VLC video player has built-in ASRC. As such, the speed can change audibly during playback. Turning off ASRC while playing local files is a feature that has been asked for years to the VLC team (it is only needed to playback realtime streaming, such as radio stations), but no one has volunteered yet to do the job. It was also used in the Reclock filter, that allowed to play back DVD in computers following the video clock instead of the audio clock, to avoid micro-judder in the picture. This one could adjust the speed more precisely than VLC, but in counterpart, it introduced clearly audible aliasing in the sound (the triangle of the orchestra in 2001 space odyssey made "tiwiwiwink" instead of "tinnnnnk").
ASRC must be avoided in high fidelity. It may be used in "anti-jitter" devices. These little boxes pretend to eliminate the jitter that comes from a CD Drive feeding an external DAC. This is nonsense. First, DACs eliminate all the jitter from their digital input. All that's left in the analog output comes from the DAC itself. Second, there is no reason for the drive to have more jitter than the anti-jitter device itself. Third, if the device uses ASRC (some may just use PLLs), it harms the sound quality much more than any amount of jitter could.

 
OK, so what is the PS Audio DLIII DAC doing when "resampling" at the DAC itself (since you define it as happening before the DAC)?
 
Thanks for the information! I'll avoid ASRC.

 

[Pio2001]

I have read the manual of the DLIII, and they say "Select the sample rate you wish. There are two choices available on the front panel: 96kHz and 192kHz. What comes out of your transport or CD player is 44.1kHz. The DLIII will upsample this to a higher, and better sounding sample rate. Choose which sample rate you wish by simply listening and deciding which sounds best on your system."
 
This is indeed resampling, not oversampling, that occurs later. The DLIII does both.
When they say "better sounding", this is just commercial. It is the same as taking a 128 kbps mp3 and converting it to the "better sounding" 320 kbps before decoding it. They probably did that to reduce the costs. It is way easier to design a driver that operates at a constant sample rate. Once the DAC is setup to run at 96 or 192 kHz for USB, I guess it is easier for them to use the same frequency for S/Pdif.
 
Although, given their advices about digital cables, that, according to them, have a "major impact" on sonic quality, not to mention power cords, that must be "shielded", it is quite possible that they really believe that upsampling 44.1 kHz to 96 kHz improves sound quality !
 
Anyway, if they do it properly, it should not decrease the sound quality.

 

[jtaylor991]

 
Anyway, if they do it properly, it should not decrease the sound quality.

 
Thanks for that! (And in hindsight, sorry for my laziness 

)
 
Key word there is properly! I wonder...
 
What got me was when someone on here said that this meant it would take a higher quality input (24/96 or 192) and downsample it on its way in, and then resample it again. Nonsense! This keeps me from really getting to try hi-res with this DAC.