If the original content was sampled at 44.1, no amount of taps can get the information back, it was tossed on the cutting room floor.
That of course assumes the combination of frequency and modulation envelope exceed nyquist.
Jn
I think we all understand that as well as Gibbs. What he said was multIple sines are transient, which is an entirely different statement.
It would be interesting to see how it gets distorted with the upper sideband eliminated.
Jn
A press brake is used for bending metal. It has a ram that has a tighter than 90 degree angle the lowers into a die. In what is called "Air Bending" the die will allow a sharper than 90 degree bend so the ram is set to bend the metal to 90 degrees and then stop its' downward movement.
The smaller and more common brake just clamps the sheet metal and then uses a ram on a hinge to fold the metal to form an angle.
My press brake can bend a 1/4" thick piece of metal 8 feet long.
It requires the operator to hold in two buttons to run. The older style used a rotating massive inertia wheel to move the ram, once started it had to complete the cycle. That resulted in lost fingers. My version has additional safeties such as dual relay circuits so that a single welded contact will not allow cycle operation.
The smaller and more common brake just clamps the sheet metal and then uses a ram on a hinge to fold the metal to form an angle.
My press brake can bend a 1/4" thick piece of metal 8 feet long.
It requires the operator to hold in two buttons to run. The older style used a rotating massive inertia wheel to move the ram, once started it had to complete the cycle. That resulted in lost fingers. My version has additional safeties such as dual relay circuits so that a single welded contact will not allow cycle operation.
No, if the content violated Nyquist for any reason, the violating information would be missing. That means that any transients surviving can be represented by sums of legal infinitely long sine waves.
If you look at how a very long FIR filter works, there are lots and lots of samples in the pipeline, maybe 20-million. Each sample in the pipeline is used to calculate current output sample. There is plenty of information about whether the correct waveform should be a continuous sine or have a fish envelope.
We can't check for infinitely long sines, but we can check for sines that are 20-million taps long. In other words, that is the degree to which we can approximate an ideal brick wall filter.
Last edited:
Since he's no longer talking to you, I think it's just semantics, I took it that he was saying there is fundamentally no difference, ie, define what you are calling a transient?
If the violating information is removed, what survives is not the same as what was originally there. And quite honestly, with it removed I have no idea what the remaining would look like, but am certain it is not identical to the original.
But if I purposely create a fish envelope modulation, how would the FIR realize that?.. Fish envelope modulation, we should copyright that..😉
Jn
Last edited:
True, it is been band limited.
It has 20-million samples to check for evidence of the infinitely long sine waves that provide the correct answer.
Hi everybody. Here is the frequency response of the B&K microphone that we used for the Greatful Dead Wall of Sound starting in 1973. It is the same microphone that was used for Mark Levinson in his recordings, George Quellet (sp), Crystal Clear Records, and I am sure several other companies. Note that it is essentially flat to 40KHz. This is the criterion that we use when designing hi end recording equipment. Many studio mikes do not meet this frequency response, but they still roll off relatively slowly, compared to the brick wall filter necessary for 44.1KHz digital. Try to learn what makes condenser microphones roll off and all will be clear.
Attachments
I believe the example of a sine modulation of 20k by 5k is perfect. Imagine total silence in the channel for days, then suddenly ping, a high amplitude blip of one lobe.
That is a transient. As is a cymbal.
A summation of multiple sines, no matter how many, won't fail as long as they are all below the limit. The organ chord is an example.
...ps he was never talking to me.
Jn
More BS. 50 year old info. Well we have learned since then. Did the 50 year old PA put out anything above 17k? ( most studio monitors back then struggled with 20k) Nobody needs a 40k BW for live sound it cause more problems than benifit, and stidio recordings arent much different. 30 years in studios with B&K mics and hardley see them used.( often refered to as sterile). Stick to talking about designing, your recording experience is obviously limited.
Last edited:
Yes, if it was a legally made fish that satisfies Nyquist in all ways. In other words, it must be able to pass through an ideal brick wall filter in an ADC unaltered. If the incoming analog waveform was a fish shape with no frequency components above Nyquist, fine.
Wow, that was pretty uncalled for. He only said what was used, and provided information.
Sheesh.
Jn
Hmmm, your sounding like me. Be careful or you may become the subject of pig-pile type ridicule.
Oh wait, you already have been....nevermind.
Jn
I should add that many recording studios found that analog tape recorders at 30IPS were better than at 15. Why, if humans are not sensitive at anything above the mythic 20kHz ?
Last edited:
Only by guys who don't know how to listen 🙂
You should see some of the PMs I get from guys who agree but don't want to take the abuse.
For the record, and in regard to how filters work, I am guilty of telling you an explanatory story making some use of license and some anthropomorphizing. I'm trying to develop your intuition a little more in a direction much closer to right than what you have been thinking. If you ever visit Auburn we can sit down and look at how FIR filters work in more detail.
I appreciate your approach. I do however have a pretty good handle on FIR and IIR filters.
My concern is with trying to reconstruct when information has been removed. Transient content has the additional issue of modulation type splash above nyquist, and I wholeheartedly object to cutting that out as it changes the output.
Jn
My concern is with trying to reconstruct when information has been removed. Transient content has the additional issue of modulation type splash above nyquist, and I wholeheartedly object to cutting that out as it changes the output.
Jn
Increased linearity of the frequency response, decreased noise (white, hiss), shifting the background (gaussian) noise spectrum toward higher frequencies (less annoying to ears), less noticeable effect of the magnetic tape drops, to name a few. Was there any recording engineer not aware of these? Nothing to do with the frequency response per se, and the comparison with the digital sampling rate is ridiculous, if not straight stupid.
Don't worry, Nyquist won't rape you.
Yet there are a few examples floating around, if I recall correctly, where test participants preferred audio reproduction that was low pass filtered to remove ultrasonic content.
- Status
- Not open for further replies.