@richardbrand - You meant to quote @phusis , not me.
Can you hear bit rate?
Almost all the music I listen to these days is from Roon and often a "station" created from an artist I like. So I click on say Melody Gardot and Roon start randomly picking similar jazz music. All great.
As Roon finds new tracks I get stuff rom Qobuz or Tidal and in a variety of bit rates. from 44.1/16 to I think 96kHz/24. Sometimes I think "wow that sounds great" and the source material is high res, other times it is not.
I've typed here for a while that around the turn of the century DAC's have gotten much better at paying Redbook (44.1/16) music than before, so that the difference in sound quality is almost gone. In addition I use Roon to upsample everything to 176 or 192 kHz.
I'm finding the question of source depth, at least with PCM, kind of irrelevant these days. What do you think?
- ...
- 52 posts total
Sorry Erik - my mistake entirely! No excuses ... I am finding this an interesting thread - thanks for starting it |
I for one have no idea who Dr. AIX is, unless you mean Dynamic Range. I found this discussion: AVS/AIX High-Resolution Audio Test: Ready, Set, Go! | AVS Forum and note that the opening post explicitly excludes DSD and SACD! DSD was explicitly designed, by the original inventors of CD, to eliminate the linearity problem inherent in PCM. It is of course startlingly easy to compare CD quality and SACD quality because almost all SACDs include a full CD layer as well as 2-channel plus 5-channel SACD versions of the same recording. You only need a system that can play these options. So what is the linearity problem with PCM? It is that each bit should contribute a precise amount to the output volume, but the contributions vary so dramatically in amount that the industry has spent 40 years coming up with improved, but not perfect, methods of doing so. I have taken numbers shown below from Audio bit depth - Wikipedia With 16-bit PCM, the range of values is −32,768 to +32,767. The most significant bit should contribute 32,768 times as much as the least significant bit. Furthermore, at the crossover point, there should be a monotonic change when the fifteen lesser bits switch off, and the most significant bit switches on. Monotonicity means that the output should always increase as you move through the number range. The numbers are even more fantastic (I mean that literally, we are in fantasy-land) with 24-bit. The range of values is −8,388,608 to +8,388,607. The most significant bit should contribute 8,388,608 times as much as the least significant bit. In reality, we probably max out around 21-bits! There’s also 32-bit. (16, 24 and 32 are convenient multiples of the now ubiquitous 8-bit byte that computers are designed around - nothing magically related to audio at all). With 32-bit, the range of values is −2,147,483,648 to +2,147,483,647. The most significant bit should contribute 2,147,483,648 times as much as the least significant bit. Wow. DSD resolves the monotonicity electrical engineering dilemma beautifully. Each bit should just nudge the output up or down by an identical amount. DSD is what I believe we should be listening to when we talk about high resolution. Caution: a lot of dacs do not natively handle DSD; DSD has to be down-converted to PCM (a lossy process) for such dacs to work |
Benchmark Media (well-known DACs and amplifiers) offers at least two white papers from their library that discuss these matters: What High-Resolution Audio is NOT - Benchmark Media Systems Audio Myth -"DSD Provides a Direct Stream from A/D to D/A" - Benchmark Media Systems
|
- 52 posts total