Resampling in DAC

[Hemicrusher]

Do most DAC's re-sample internally to their native resolution? I was reading about the pros and cons of re-sampling in software and sending to the DAC at the native resolution, but opinions were all over the place. Just curious if all DACs do it, or not.

Thanks.

 

[yue]

Expensive DACs will have a few different crystal oscillators to support different clock rates. For instance, 45.1584Mhz and 49.152 Mhz, in order to support 44.1khz (and its multiples) and 48Khz (and its multiples) natively. So for those DACs, they have multiple native resolutions.

Others will process the PCM data to the only native clock rate available in the hardware, so may have higher jitter issue. For instance, the ODAC implementation only has a 12 Mhz clock, so both 44.1khz (and its multiples) and 48Khz (and its multiples) are converted to that rate. In theory 44.1khz may see a little bit higher jitter since the data processing may be slightly off phase. But if done well, the noise should be inaudible.

 

[Blumlein 88]

I wonder if what you've read is about feeding sigma-delta DACs. Eventually those run at a very elevated sample rate using quite often 5 bits. It is true many good DACs have two clocks for multiples of 44 and 48 khz. If you feed sigma-delta DACs 44 khz they may resample to 8 times that and then resample again to the internal native rate. If you feed at the highest rate of say 192khz or 384 khz, it may resample only once internally. So the idea is to resample as close to the actual internal rate to have less resampling going on.

It seems like a nice argument I suppose, but I've seen no measurements to indicate one way results in better performance at the analog outputs of the DAC vs the other way of feeding the actual rate of the music file.

 

[yummy]

that's called oversampling, of course it should! 256fs, 512fs, 1024fs, higher means crazier

 

[Hemicrusher]

I downloaded Audirvana for Windows and noticed that it sends 16/44 files to the DAC at 24/44. Now I am new to this and the 24bit part is the resolution? The developer says that is because the DAC can only accept 24bit files which is why it shows. Other players just hide this, since it has zero impact on the file.

So, is this how DACs work, that they can only accept one type of bit depth, yet multiple sample rates? (I am sure I have the lingo wrong)

 

[Soniclife]

So, is this how DACs work, that they can only accept one type of bit depth, yet multiple sample rates?

Most DACs support multiple sample rates and bit depths, it's common to see 44/16 & 44/24 listed being reported by the driver as to what it supports. However there is no issue with sending 16 bit data to a 24 bit DAC, it should sound the same. Not supporting 16 bits sounds like a firmware or driver issue though.

 

[bennetng]

I downloaded Audirvana for Windows and noticed that it sends 16/44 files to the DAC at 24/44. Now I am new to this and the 24bit part is the resolution? The developer says that is because the DAC can only accept 24bit files which is why it shows. Other players just hide this, since it has zero impact on the file.

So, is this how DACs work, that they can only accept one type of bit depth, yet multiple sample rates? (I am sure I have the lingo wrong)

No a complete apple to apple comparison but think audio bit depth as video bit depth. You don't need to set the GPU to output 8-bit when viewing some 8-bit gifs since the 24 or 32-bit palette includes all colors in the 8-bit palette.

 

[Hemicrusher]

No a complete apple to apple comparison but think audio bit depth as video bit depth. You don't need to set the GPU to output 8-bit when viewing some 8-bit gifs since the 24 or 32-bit palette includes all colors in the 8-bit palette.

That is the way I understood bit depth. Like looking at a blurry photograph with better reading glasses. Still blurry.

 

[yue]

I downloaded Audirvana for Windows and noticed that it sends 16/44 files to the DAC at 24/44. Now I am new to this and the 24bit part is the resolution? The developer says that is because the DAC can only accept 24bit files which is why it shows. Other players just hide this, since it has zero impact on the file.

So, is this how DACs work, that they can only accept one type of bit depth, yet multiple sample rates? (I am sure I have the lingo wrong)

If dac does not support 24, software can pad eight 0s behind 16bits, and send it to the dac.


using higher bit rate is always encouraged, as it leaves room for software volume adjustment. a 24 or 32 bit depth will be lossless for 16 bit file, even with a very low software volume.