And you think that Scott was confused by seeing the spectral envelope of the decaying 5Khz sine wave, or that he was he confused because he saw that the spectrum was widened ?
Or do you think he was confused that when brick wall filtering at 5.1 Khz that both the time domain signal and the frequency domain spectrum will get changed ?
Hans
I don’t agree with you regarding the raised noise floor.
Here’s a link with many references to other (AES) papers, covering the subject quite well.
A Case of the Jitters | Stereophile.com
Hans
Darn it, I'll have to look Monday. Two books come to mind, both digital signal processing books. The discussions on envelope modulation come to mind.
The 5 kHz modulated example I put up showed the spectra with a spread. That spread is actually the spectra of the modulating waveform, the exponential.
The problem with an FFT is that it is an integrated total over a time. It cannot show how the spectra changes at any instant. An exponential decay for example, has a different spectra at every instant in time.
If you remove some of the USB content, the FFT will show the truncation, but it does not tell you when in the waveform it is removed.
That is the beauty of Scott's method of using the difference to spot when the filter pulls content. And by when, I mean instantaneously.
My specific mention of a time mirrored exponential modulation, that is a signal which will have it's sidebands spreading out as the envelope slope rises. Scott's difference test will not remove any content of the wave until the modulation slope is steep enough that the USB reaches the Fs of the brickwall.
Jn
Last edited:
So as I understand it, the microphone is the recording engineer's (as artist) "paintbrush", so what we hear is as much the product of that art, as it is the musician's output. Particularly in the multi-mic'd / mixed down studio recordings where (I assume) the engineer's goal is to make it sound good on playback across as many devices as possible, from the dishwasher's boombox to a high end system.
Contrast this to an idea such as a minimally mic'd recording of musicians in a room - where tracks 19 and 20 are the room impulse for left and right, where one could unravel the room conflaguration and put the musicians in your own room.
I'm still intrigued by the GD wall, where it was intended for the musicians to hear just what the audience hears - instead of hearing what comes out floor monitors while the audience hears something else. Quite a spectacular attempt at giving the musicians more control over what the audience hears and I can read how a lot of work went into expressing that idea. Still curious "what happened"? Perhaps musicians (generally) just want to perform - and leave the details of how it sounds to somebody else...
Also, what do listeners prefer - a minimally mic'd recording - or - one where the engineer is free to use certain mics for certain parts of an acoustic instrument, angle them in some particular way and mix down so that it sounds "as-if" the ensemble is "right there" in between your stereo speakers. Which I assume is the usual way its done. So whomever has their craft the most polished should sound the best - maybe its time to uplevel from the individual hardware level of discussion to who's doing the best sounding work as a music recording artist.
I once rapidly approached Dave Grusin during a local performance break, I think I made him nervous. I thanked him for his GRP digital master studio work and commented that it always makes the stereo sound great. He said "well, we want it to sound good too!" as I shook his hand. I told him its great to be here tonight and then vanished so someone else could have a turn.
I havent been on stage in 20 years (bass) and from what I remember it was all about holding up the band for the song - my little brain had no room for small details in how it sounded. Perhaps when you're really good...you can think about that and play at the same time. Maybe even then its preferred to simply leave those to somebody else you trust.
Contrast this to an idea such as a minimally mic'd recording of musicians in a room - where tracks 19 and 20 are the room impulse for left and right, where one could unravel the room conflaguration and put the musicians in your own room.
I'm still intrigued by the GD wall, where it was intended for the musicians to hear just what the audience hears - instead of hearing what comes out floor monitors while the audience hears something else. Quite a spectacular attempt at giving the musicians more control over what the audience hears and I can read how a lot of work went into expressing that idea. Still curious "what happened"? Perhaps musicians (generally) just want to perform - and leave the details of how it sounds to somebody else...
Also, what do listeners prefer - a minimally mic'd recording - or - one where the engineer is free to use certain mics for certain parts of an acoustic instrument, angle them in some particular way and mix down so that it sounds "as-if" the ensemble is "right there" in between your stereo speakers. Which I assume is the usual way its done. So whomever has their craft the most polished should sound the best - maybe its time to uplevel from the individual hardware level of discussion to who's doing the best sounding work as a music recording artist.
I once rapidly approached Dave Grusin during a local performance break, I think I made him nervous. I thanked him for his GRP digital master studio work and commented that it always makes the stereo sound great. He said "well, we want it to sound good too!" as I shook his hand. I told him its great to be here tonight and then vanished so someone else could have a turn.
I havent been on stage in 20 years (bass) and from what I remember it was all about holding up the band for the song - my little brain had no room for small details in how it sounded. Perhaps when you're really good...you can think about that and play at the same time. Maybe even then its preferred to simply leave those to somebody else you trust.
Last edited:
Thanks, no hurry. I think I'm getting most of what you are describing but I think some visual aids would help me more (animation would be excellent!) Hopefully your references will have pictures 🙂 "Digital processing books" ok...digital is obviously not one of my best understood areas 🙄
A college textbook from the 70's with visual aids or animation? 😕
We are all spoiled by Google...I am no exception. Googling may actually be better for you than my old texts. Wiki does tend to be fairly accurate now.
I have not found yet any treatise which shows how instantaneous frequency evolves out of a modulation in a graphical or animated way. Hints do show up on occasion, even here on forum. Like how, when a sine is instantly turned on, the start violates nyquist even though the sine itself doesn't. That is actually modulation with a square wave, and the leading edge of a square wave is the violator.
Jn
Focusing on this bit, it seems to me an instant in time cannot have a spectra as such, unless I'm missing the point entirely this is where the confusion over "frequency shift" comes in, it's actually a phase/time shift which could cause the distortion?
For my example of the time mirrored exponential modulation, visually you cannot tell what the sidebands are doing. Using Scott's brickwall/difference stuff, you will see when the filter reacts to the sideband content exceeding Fs.
I had been toying with the idea of a phase locked loop to see instantaneous frequency, but the ones I use at work rely on a constant amplitude. I have no idea how to adapt one to a time varying envelope.
Oh btw, I found a wiki that went through the beat frequency math, and at the very end of the section, it said if the mix is an audio signal, we will hear the carrier frequency, not each individual one. It cited a reference, I will try to reaquire that link.
Done..
Tipler and mosca, physics for scientists and engineers..vol 1
The wiki page I got to by googling envelope modulation, found envelope(waves), can't seem to copy the link on my iPad.
Last edited:
I did setup a convolution in excel, but tried only sin(x)/x. I didn't try any of the wavelets I found using Google. Results didn't float my boat.
Jn
This got my attention.... is there a some what normal condition with music transients which is similar .... start violates nyquest?
THx-RNMarsh