Roon's 64 bit floating point volume leveling

No volume leveling is lossless, this is absolutely correct. You always loose resolution no matter how you do volume leveling. This is why listening to loud music gives much more pleasure than quiet music.

Loss includes analog volume leveling! Fact is: it produces much more errors (in the measuring domain) like noise or distortions as a well done digital volume control.

I perfectly understand that. But from a certain point of view it seems that leveling on 64 bits is less lossy than leveling on 24 bits (even if the ear resolution in under 24 bits). So in the roon’s signal path colors, they’ve made the distinction between those two cases - which make sense (at least for me).

I haven’t tried yet the roon’s DSP volume on 24 bits. All I know so far with digital volume control is the leedh processing in my 3D lab streamer which work very very well.

I will soon be able to compare Leedh processing vs. Meitner Vcontrol vs. Roon DSP - My ears will judge :slight_smile:

I’ve read that roon bridge DSP volume attenuation is the exact same code than in the core : so it means a 64 bits conversion before applying attenuation, then a conversion back to the original bit depth.
Is that correct ?

It also means that the color code for the signal path is indicative but does not reflect absolute reality.

If we have a volume leveling at core level and then volume control at bridge level, it means that the conversion to 64 bits will happen twice, and so does he dithering. Don’t you think it can be audible ?

I understand the technical choice that roon has made. I hope that the sound quality remains “magical” even with all those treatments. Because I know it is possible to have a really good digital level control. And in that case, a DAC directly plugged in the amplifier (no preamp) is amazing.

1 Like