DACs and aliasing

[r010159]

I have two DACs. I have been playing the same material, first on one DAC, and then the other. One DAC has a type of distortion, kind of an inharmonious quality to the sound in the highest register of a singing voice. The other DAC does not have this problem. I did check for clipping but where this happened had no evidence of clipping.

 

Could this be aliasing? How would this be possible for a DAC? I thought oversampling and the reconstruction filter takes care of this? Can there be a difference in how one DAC reconstructs the analog signal compared with another DAC?

 

Thanks!

 

Bob

 

​PS: I wonder what aliasing actually sounds like?

[Miska]

Could this be aliasing? How would this be possible for a DAC? I thought oversampling and the reconstruction filter takes care of this? Can there be a difference in how one DAC reconstructs the analog signal compared with another DAC?

 

Technically it cannot be aliasing in DAC, unless it converts between sample rate families in which case it is possible if the digital filter is not good enough. (aliasing here means higher frequencies folding down to base band)

 

However, DAC needs to have anti-imaging (reconstruction) filter to remove images (sometimes also called aliases) outside base band. This is typically combination of digital and analog filters. If this filter is not strong enough, these high frequency images can cause intermodulation distortion, since in many cases non-linearity in amplifier stages inside and outside the DAC increase as function of frequency.

 

Knowing the particular DACs could help determining which one could be the case.

 

PS: I wonder what aliasing actually sounds like?

 

It is similar to what you've described, but it also could be IMD. I've made some sample tracks of how aliasing sounds, I'll try to make those available for download later today...

[Skeptic]

Could it be intersample clipping? Try reducing the playback volume in the PC by 6dB, or analyze the tracks with JRiver's analyzer and it will tell you if there's clipping.

[r010159]

The DACs in question are the O2/ODAC and the Apogee Duet 2. I am getting that artifice with the ODAC.

 

Bob

[buonassi]

i am curious how "aliasing" sounds.  what do I listen for?  I upsample 44.1 to 352.8 and even when I use a very weak filter in izotope, that I know is not suppressing any of the images above nyquist, I can't hear a difference?  I have pretty good hearing (or so my ego caused me to type).  I can hear ringing, especially pre-ringing. So a few questions:

 

1. is it that my equipment doesn't reproduce frequencies higher than nyquist, and therefore doesn't contribute IMD? is it possible there is an analogue filter stage in my dac or headamp that is filtering out frequencies above 20K let's say?
2. is aliasing an issue only when downsampling and not upsampling?
3. is it my headphones not being resolving enough?  I have a pretty respectable setup I think.  
4. Is it my hearing?  I'm just not trained to hear these things and they are very subtle differences in sound?  I imagine that aliasing would sound as bad in the audio realm as it looks in the video realm!  But perhaps the tells are harder to pick up on?

 

 

 

[bogi]

On 5/28/2014 at 8:52 AM, Miska said:

I've made some sample tracks of how aliasing sounds, I'll try to make those available for download later today...

 

I'm interested!