The statements are actually quite correct.
However, that text as written is correct for a digital system such as a PID loop, where the feedback and computations are delayed by a time interval. For example, the systems I am working with have a 400 microsecond servo update rate. I can see this also impacting an FIR and IIR scheme as well.
To blindly assert that occurs in an analog system as stated, well...pfft.
jn
Understood. I used ASRC as an example primarily because my reading is bit ahead in that area right now relative to where it is for S-D modulators. However, it turns out that S-D modulators are also asynchronous oscillators in some sense even though S-D dacs benefit from ultra-low phase noise clocking. Not prepared to make a leap into talking about that yet.
Regarding Scott's point to me about FFT repeatability vs ASRC, I see the point, but as is often the case certain details can matter. I would agree that offline standalone FFT should give the same results every time. In the same sense, rational fraction sample rate conversion of a saved file should give the same answer each time (although it is or may be a numerical approximation carried out to arbitrary accuracy). Where things get more complicated is in approximately real-time systems. When doing polyphase interpolation to much higher output sample rates all at once, there is a lot of zero stuffing that takes place. As a result, many FIR calculations will be multiplies by zero, which can be skipped since the answer is already known. Where errors tend to crop up is in real-time ASRC where the input sample rate has to be tracked by a VCO to help pick the right pre-calculated coefficients for each sample calculation. In that case one source of VCO error is jitter in the input stream that the VCO needs to track. For small time interval errors due to jitter, they can be equated as about the same as small bit errors in sample data in terms of the effects on resulting calculation accuracy. For the case of a system that performs real-time FFTs from incoming analog data, clock jitter in the ADC and noise present in a nominally repetitive input waveform may cause FFT calculations to vary as a function of time. Considering all the above, FFT vs ASRC might seem a bit closer to analogous in some way if comparing in an apples to apples type of context. Despite having said the foregoing, I accept that FFT may be more common in an offline context as opposed to ASRC which kind of implies a real-time system.
As an aside, as Bruno Putzeys has been mentioned recently, perhaps a good time to share the thought that it may turn out his earlier work with dacs and understanding how they work is not unrelated to understanding free running oscillators in the context of SM audio amplifiers. 🙂
Last edited:
Steely Dan - Aja. Latest remastered version is not the most popular with some, but the tails are there especially for vocals.
Dang, I only have that on vinyl 🙂
Surely choral music would be good for this. Nothing like a cathedral for an 8 second decay?
Surely choral music would be good for this. Nothing like a cathedral for an 8 second decay?
well said and explained. 🙂 😎
THx - Richard
One could use a single exponentially damped sine wave as the reverb tail signal. See jam.. I gave him my HP arb waveform gen. or do it in software.
Last edited:
That’s what they said, you know, the marketing department. It is obviously false if you have any understanding of it. It is no more fit for archival than PCM.
tthey thought if the coding standards changed in the future --- number bits and formats etc... they could convert easily from DSD to any other popular format. Not a marketing decision.
THx-RNMarsh
Last edited:
This is all well understood stuff, especially outside audio.
You state the FFT is more offline but that is not true. There are billions of devices doing signal processing in the frequency domain in “real time”.
Why is that true? Why is it not true for PCM?
You should understand before you repeat their sales material again.
No, actually I did not state that. I said FFT may be more common in an offline context.
Which is also probably wrong given the prevalence of cheap DFT and filtering in the frequency domain.
Do we really need diatribes with no line breaks that discuss DSP like it’s just been invented in your quest to justify your story regarding jitter and reverb tails?
You can just ask Benchmark, but they’ll probably say your unit is defective if you hear this. Or maybe they will just laugh, I don’t know.
Last edited:
No.
I just do as JC said. I just listened. That .04% was the HPA I bought 4 years ago and didnt like which led me to make my own.
I had never measured it until yesterday to see what the thd is with my load Z.
THx-RNMarsh
Last edited:
I dont know why and dont care to argue with you.
Thats their story. ask them.
THx-RNMarsh
Others already have done ABX of DACs as old as late 90's - early 2000's production which you can read about. They all sounded excellent beyond our hearing ability. It's a matured technology. FYI, if you are really into improving reproduced sound, speakers and room acoustics are the areas to put your time and effort into. But then if you aren't, anything goes. Say, ASI resonator?
I don’t need to ask, since I already know the answer and gave it to you. I was just hoping you’d do a quick search so you could figure out why rather than parroting their narrative which isn’t based in fact.
DSD is essentially a dead or nearly dead format that offers no real advantage over 24/192 PCM. It looks even more pointless when you consider that 99% of recordings will have to be converted to PCM for any sort of processing and then back for pressing onto the SACD or however you want to distribute it as DSD.
Last edited:
OK, I would never have thought a top SOTA playback system would not buffer this out.
- Status
- Not open for further replies.