Thanks Hans!
The filter has zero of any hocus-pocus apodizing, anti pre-ringing, assymetrical jumbo-bumbo whatever. Just Nykvist-Channon et al. compliant technology.
+ conservative use the BW and no aim for 0,00001 dB down with 0,0000000001 dB ripple upto 22,03 kHz.
//
Not necessarily - a mono recording (it was originally intended for dynamics and spectrum measurement) could easily be duplicated to create a stereo set with varying ITD to see what's audible?
Of course we can use the same maths that we are trying to prove do not work to create the sub-sample delay. 😉
I always gives me pause that these folks were still around after I had finished school and was working in the industry they helped establish. Claude Shannon's daughter was my neighbor in Cambridge for a few years, we had group get together and she introduced herself and I said, "You must be Claude's daughter" the resemblance was so strong.
Who cares what ITD of cymbals is audible? If it isn't use ILD like they do already 🙂 Anyway Jn wants to measure it first.
If sub-sample delays were not possible, then we couldn't reproduce an infinite range of frequencies below 20kHz. Both things require zero crossings (or any other point on a wave) between samples to be possible. The whole discussion makes no sense.
Whether or not humans can hear ITD at HF is another question entirely.
EDIT: In addition, it has been stated multiple times by various people that it takes a long time to reconstruct frequencies near Nyquist, it means we need a long filter to do it with high accuracy. How long? How accurate do you want?
Seems like JN's question, since he asked about proof of linearity, is whether or not when f is close to Nyquist and more than one frequency is present, do the frequencies non-linearly phase shift one another? Is that it, or is it something else JN?
Last edited:
Don't forget what goes to your amplifier/speakers is a continuous analog waveform, the concept of "between" samples is rather nebulous. I'm afraid in this case the burden of proof is on jn the failure of the theory here borders on extraordinary claims.
Last edited:
True. We could look at samples and software waveform after the samples have passed through a reconstruction filter. If we know the time delay introduced by the soft reconstructor, we should be able to line up the sample points with the reconstructed waveform they produced. However, the accuracy of the soft analog waveform depends on the particular filter performance.
Last edited:
Regarding luxury, people sensitive to this kind of elegance do not tinker compliments themselves, but leave this care to others.I'm afraid you cannot afford the luxury of arguing or being "at war" with yours truly. Carry on with your ignore list.
And there is a saying that culture is like jam, the less one have, the more he spread it. You are, even in your small specialty, a kind of king of the thin layer.
Which it is all the more difficult to hang on as your sense of humor that is intended to serve as a support seems to have the density of the vacuum.
With my compliments.
Last edited:
As most of the recordings/mixs of modern music are done in close miking and localisation by "pan" on the mixing desk that use only level differences, what about ITD importance ?
Last edited:
There's a new listening test poll open regarding cymbals, head over hear and take part.
Poll: Cymbals of different sampling rates listening test
Cheers Michael
Poll: Cymbals of different sampling rates listening test
Cheers Michael
Yes, I wish all consumer products actually did so well. It is what you expect. But it also needs to be built and measured to see how well the SIM can be done especially above 20KHz to HF as you appear to have done. Is that what you did? Built?
Can you write up a DIYAudio circuit for us to build and see what happens in practice with various available parts?
THx-RNMarsh
Last edited:
I didn't see this answered. There's two overlapping meanings of DSP.
Any CPU can do Digital Signal Processing, but a Digital Signal Processor has special architecture (such as Harvard, separate program and data busses) and special instructions (such as multiply-accumulate or MAC) to do signal processing faster. On the other hand, many general-purpose CPUs have included some of the special instructions to speed up DSP operations on those CPUs.
Digital signal processor - Wikipedia
Thank you. I try.
After you get the filtering sorted and no HF artifacts And, there seems to be others outside DIYAudio doing similar tests with cymbals and all because the sound falls apart above midrange (not accurate sound). Another issue with CD; sampling, GD issues?
Next is to answer my question regarding the 50usec one-shot pulse signal ... a non-continuous signal.
After that, then I can always think of more for you to do...
Stay busy in 2020.
THx-RNMarsh
Last edited:
One ought to clearly distinguish one make/model from another. As you could in live listening.
THx-RNMarsh
OK I asked to see this in practice. How to get it and use it. Other-wise not sure what the point is other than another method for a LPF albeit less hocus-pocus. Maybe less artifacts? so it appears. How can others get it in 'canned' form and use it?
I, personally, am not going to do anything with it but others could. More details and info, pls.
-RNM
Last edited:
Get a DSP dev board (Sharc or Blackfin from Analog Devices, or an OMAP from TI). Ask one of your EEs to program it (not really rocket science, everything he would need is in libraries). Here you go.
Using what to do the recording? Having just a recording is only part of the avaluation. How to compare real vs recorded when real isnt in the room to compare against (other's homes)? Unless, you and Mark with his system compare and judge the repo accuracy for that repo system?
THx-RNMarsh
Last edited:
- Status
- Not open for further replies.