T,
Me tinks you are stretching the rubber band on this a bit too tight, no one is advocating high distortion anywhere, in fact the contrary.....It is not possible to achieve the low levels that is achievable with electronics, one is electro-mechanical distortion and the other is'nt.
The ears is more sensitive to the electronics......
Me tinks you are stretching the rubber band on this a bit too tight, no one is advocating high distortion anywhere, in fact the contrary.....It is not possible to achieve the low levels that is achievable with electronics, one is electro-mechanical distortion and the other is'nt.
The ears is more sensitive to the electronics......
Hi,
If electonic and electromechanical distortion follow the same transfer function, how are they different?
Really, if the same transfer function is produced by electronic and by electromechanical means, how will yours tell which is which?
This is WAY up on my Voodoo meter. The human ear can tell if the same distortion is produced electro-mechanically or electronically AND is more sensitive to one kind than the other. Wow, I'm impressed. I did not know that. Must be ESP. Provide proof and claim the 1 Million from Randi and very likely a Nobel into the bargain.
Ciao T
If electonic and electromechanical distortion follow the same transfer function, how are they different?
Really, if the same transfer function is produced by electronic and by electromechanical means, how will yours tell which is which?
This is WAY up on my Voodoo meter. The human ear can tell if the same distortion is produced electro-mechanically or electronically AND is more sensitive to one kind than the other. Wow, I'm impressed. I did not know that. Must be ESP. Provide proof and claim the 1 Million from Randi and very likely a Nobel into the bargain.
Ciao T
Sure, if you know what you're looking for. But you'd not look at the entire 4:32, you'd be interested in the spectral composition of the finger squeaks and probably gate the measurement around them. For what reason, I have no idea- as a tool, spectral analysis is useful for characterizing the performance of electronics and transducers, so it might be better to get them working as well as possible using various controlled test signals, and understand that if you get them right, the squeaks will sound like squeaks, not like thunderclaps.
I'm not sure what you mean by "rounding error" in this context- most calculations are done at a much higher bit depth than the recordings. And when you say "compare signal waveforms," I assume you understand that signals can be represented equally in the time domain and frequency domain?
Because nobody yet have model of "comparator" used by our auditory perception.
Take several microphones and place close to each other. All of them will give you different vaweforms, while ears placed on their position will register still the same sounds of finger on strings. What are you going to compare?
Last edited:
Electronics is prone to generate such a "strange" for hearing kinds of distortions, that an electromechanical system is not capable to generate in principle.
Therefore ears are sensitive to specific distortions from electronics, but these distortions are not reflected by THD, maybe transient response is closer to truth.
Vladimir,
Yes.
You may say, that what I am attempting is to change the nature of the electronic distortion and to make it more like electromechanical distortion in nature.
The price to pay for this is to accept more distortion overall, not that it matters.
Ciao T
Yes.
You may say, that what I am attempting is to change the nature of the electronic distortion and to make it more like electromechanical distortion in nature.
The price to pay for this is to accept more distortion overall, not that it matters.
Ciao T
Well you must have special voodoo T ... I hope your amplifiers have a better squarewave response than your speakers, same TF ..?
From the Fourier analysis wiki: "In signal processing terms, a function (of time) is a representation of a signal with perfect time resolution, but no frequency information, while the Fourier transform has perfect frequency resolution, but no time information.
As alternatives to the Fourier transform, in time–frequency analysis, one uses time–frequency transforms to represent signals in a form that has some time information and some frequency information – by the uncertainty principle, there is a trade-off between these. These can be generalizations of the Fourier transform, such as the short-time Fourier transform, the Gabor transform or fractional Fourier transform, or can use different functions to represent signals, as in wavelet transforms and chirplet transforms, with the wavelet analog of the (continuous) Fourier transform being the continuous wavelet transform."
Fourier as many know was concerned with solutions of heat problems (some history of him and his predecessors in the above wiki). The usefulness of the maths for periodic functions of time is well-established, and the extension to aperiodic a bit less well-understood, but still useful. But --- for "perfect" accuracy it is a bit of shoehorning phenomena into a very long idealized series of integrals.
But why does it work as well as it does? One can argue that it's because the ear/brain/mind listens for periodic, or quasiperiodic, acoustical signals. And most music contains these to some extent, even if the nuances are considerably more complex than what can be efficiently represented by sinusoids.
Coming at things from another angle, why haven't wavelets displaced Fourier analysis? They allow a systematic tradeoff of time resolution and frequency-like resolution. But most of the time music doesn't need this so much.
As far as the use of sinusoids to characterize the performance of nonlinear systems like amplifiers and loudspeakers, they are useful for revealing gross nonlinearites. They don't readily reveal some other departures from things behaving like time-invariant linear operators. Whether these are at the root of the disparity between listening preferences and simple distortion spectra is, I think, still an open question.
Brad
Theoretically, the frequency domain is equivalent to the time domain. But any book on numerical calculations says that even with 128 bit word length the transfer from time to digital domain is not perfect. I personally did quite a lot of complicated numerical simulations, including molecular dynamics simulations of the processes in the near-surface layer of Si under ion beam doping by various atomic species. Changing of even the last bit in initial conditions leads to completely different trajectories of atomic particles. The same will be with Fourier transform of long enough waveform.
No, it's not perfect. It's good to 128 bits. That's considerably better than you need for characterizing finger noises on a guitar.
The calculation you describe sounds like a nonlinear dynamics problem in QM, which is not terribly relevant. Unless you're suggesting that guitars are quantum mechanical.
I've had a fair amount of experience in both molecular dynamical calculations and acoustic measurements and don't see the connection at all. Maybe that's why a speaker frequency response looks the same whether you have a 32 or 64 bit OS in the computer performing the FT.
The calculation you describe sounds like a nonlinear dynamics problem in QM, which is not terribly relevant. Unless you're suggesting that guitars are quantum mechanical.
I've had a fair amount of experience in both molecular dynamical calculations and acoustic measurements and don't see the connection at all. Maybe that's why a speaker frequency response looks the same whether you have a 32 or 64 bit OS in the computer performing the FT.
Stu, it is not about "does frequency response of speaker look the same". It is about, "Does reproduced sound fool imagination as if it is real".
No matter what I measured and how I measured, I still go almost blindly, even though have some clues about what is more significant, what is less in terms of distortions. Trials and errors, trials and errors...
No matter what I measured and how I measured, I still go almost blindly, even though have some clues about what is more significant, what is less in terms of distortions. Trials and errors, trials and errors...
I have lost count of the number of discussions of this type I have seen, though I must say, rarely and juicy as the one here and now.
I have heard amps measureing near perfect, yet sounding dull and uniteresting.
I have heard amps measuring comperatively large distortions of all kinds, yet sounding really nice.
In the end, it seems to me that I have true high fidelity if a trained musician can sit down and say: "That's a Les Paul, not a Gibson guitar.", or "That's a Yamaha, not a Steinway piano". If say 7 out 10 musicians can do that, and agree on what's what, I feel I hit the jackpot.
That's when I stop, because I feel I have nowhere else to go any further. I also feel my time will be MUCH better spent on enjoying the music collection I have, than on inventingy sexy new circuits to accommodate each and every theoretical viewpoint there is, because there sure are plenty.
There's too much theory floating around, it seems to me, and too little practice. Many can sit back and pick faults in other peoples' circuits, and the odd thing is, many a time criticisms are made by people who have not even seen the unit, let alone auditioned it. But they know from theory it can't be good - how is this less fundamentalist than those who feel that if a cable is twisted to the right, it can never work as well as when it's twisted to the left?
I always though theory, i.e. science, was invented to result in something practically useful? Not to be an art unto itself, for its own sake.
So gentlemen, let's see some of your works, even if they are only simulator circuits at this point. I would like to learn from those who know better and a picture (schematic) is worth a thousand words (papers).
I have heard amps measureing near perfect, yet sounding dull and uniteresting.
I have heard amps measuring comperatively large distortions of all kinds, yet sounding really nice.
In the end, it seems to me that I have true high fidelity if a trained musician can sit down and say: "That's a Les Paul, not a Gibson guitar.", or "That's a Yamaha, not a Steinway piano". If say 7 out 10 musicians can do that, and agree on what's what, I feel I hit the jackpot.
That's when I stop, because I feel I have nowhere else to go any further. I also feel my time will be MUCH better spent on enjoying the music collection I have, than on inventingy sexy new circuits to accommodate each and every theoretical viewpoint there is, because there sure are plenty.
There's too much theory floating around, it seems to me, and too little practice. Many can sit back and pick faults in other peoples' circuits, and the odd thing is, many a time criticisms are made by people who have not even seen the unit, let alone auditioned it. But they know from theory it can't be good - how is this less fundamentalist than those who feel that if a cable is twisted to the right, it can never work as well as when it's twisted to the left?
I always though theory, i.e. science, was invented to result in something practically useful? Not to be an art unto itself, for its own sake.
So gentlemen, let's see some of your works, even if they are only simulator circuits at this point. I would like to learn from those who know better and a picture (schematic) is worth a thousand words (papers).
No, it's not perfect. It's good to 128 bits. That's considerably better than you need for characterizing finger noises on a guitar.
The calculation you describe sounds like a nonlinear dynamics problem in QM, which is not terribly relevant. Unless you're suggesting that guitars are quantum mechanical.
With all respect, measurements of speaker frequency responce do not include inversion of matrix 10 million per 10 million size (4min x 60sec x 44100 1/sec registered values during 4 min recording). Amount of operating memory needed for only storing the initial matrix with 128 bit accuracy will be 10**7 x 10**7 x 16 byte = 1,6 x 10**15 bytes.
I show these nonsense numbers just to illustrate that theoretical and practical things can not be equal, and nobody still provided proofs that low-level signal information is catched by a routine software.
Well, I didn't say anything about "frequency response of speakers look[ing] the same," (as what?) so I'm not sure what point you're trying to make.
No, they don't. Why would they? I was just responding to your incorrect assertion that FTs can only be performed on periodic signals. Speaker measurements are routinely performed with impulse and MLS, as well as periodic stimuli.
I have no disagreement with that. We were talking about a side issue, the purported limitations of one particular measurement tool, albeit an incredibly useful one.
- Status
- Not open for further replies.