You may be confusing 'can we calculate it exactly?' with 'does it exist; is it true?'.Jakob2 said:
We may want to know this but the facts are that we cannot know this because the question you ask is meaningless. Only an infinitely long sine wave has an exact frequency. Only an infinitely wideband signal has a definite time. That is, our problem is not that we cannot measure time and frequency simultaneously to any degree of accuracy but that they do not simultaneously exist so we have to choose one or the other or some suitable combination of both.mmerrill99 said:
Perhaps what you meant to say is that we often want to know the approximate frequency at an approximate time? That we can know.
At a point in time all that 'arrives' is a single pressure. Take a period of time and what arrives can be regarded as a time series of pressures, or as a set of sine waves. My understanding is that the ear works mostly in the frequency domain in the mid and higher frequencies and mostly in the time domain at lower frequencies. As far as the ear is concerned most of music is perceived as sine waves; it knows nothing of a pressure wave unless bass instruments are involved.mmerrill99 said:
My understanding is the opposite of this, but I do not claim to be an expert on these matters. You are telling us that all the stuff about frequency-sensitive 'hairs' in our ear which we read about is untrue? Anyway, exactly how our ears work is not relevant to the fact that a pressure waveform (in the time domain) is exactly equivalent to pressure sine waves (in the frequency domain). This is a fact, which you still appear to be denying - for reasons which are unclear.
Oh, don't mention models or theories on here or the tiki torch brigade will reappear 😉
I think it's generally agreed that a combination of top-down & bottom-up are in play so we are constantly evaluating the incoming signals Vs the current internal but dynamic model of the auditory scene. So prediction of what sound is to be expected next forms a large part of how we hear & if a loud, unexpected noise is heard, a mismatch negativity reaction is formed.
I often think that great composers create this unexpected combination of follow on chords, notes & it can thrill us rather than shock us.
A bit like art allows us to see a new way of looking at the everyday world - the artist's vision - it gives a different perspective on the everyday & brings joy
For sure but it is the same problem talking about audio electronics - some just want to stay in the realm of engineering & ignore the fact that the goal of audio electronics & this hobby is to connect to music in a way that thrills & interests - not because it uses technicolour audio tricks which initially please but because the reproduced music is much more realistically portrayed. In other words auditory perception is the final judge of the reproduction
Recreating all the notes in the right place is just the basics - something else is at play & we have to turn to the workings of auditory perceptions to try to tease out what it is
If you knew anything about Technicolor (sic) you would not have said any of that.
Please do not use the term FFT as a synonym for Fourier analysis; it creates confusion, and may be a sign of confusion. Fourier theory tells us what is and tells us how to calculate it; it still is whether we calculate it or not. If that is what others call "mathematical realism" then I am happy to plead guilty, although I would expect anyone who knows enough mathematics to agree that a continuous waveform and its Fourier decomposition are fully equivalent views of the same phenomenon whatever philosophy of mathematics they subscribe to.mmerrill99 said:
I'll assume that when you say FFT you actually mean Fourier. Yes, it is true that a Fourier analysis is not the only reality; the original time domain waveform is also reality. They are both real in exactly the same sense, but some of us seem to wrongly perceive that the time domain view is somehow more real than the frequency domain view.
If our early experiences in life were mainly frequency domain instead of time domain we might be having the opposite argument: you would be saying that the sine waves were the only reality, that a time-based waveform was just a model, and that our ears do not use Inverse FFT.
Thank you for a clear post. As you say, it depends what we are talking about. Perception of sound involves many unknowns, but what sound waves are is clearly understood. To paraphrase(?) mmerrill, "sound(sine) waves don't care what goes on in our ears and brain" 😉
Good god, man will you get over your extreme bias against me, trying to prove that every word I say is wrong. Everybody knows what was meant by technicolor, FFS?
MM, I understood your usage of Technicolour as an expression.
ToS, please explain what you mean by your comment so that we all may learn.
Dan.
So this Fourier transform is a deconstruction of a complex wave to its pure sine wave components?
If so is it something that can be measured real time or does it only exist mathematically?
If so is it something that can be measured real time or does it only exist mathematically?
Dan,
Thank you for asking, certainly appreciated. I’m busy at work in the studio right now, and will sit down and write about Technicolor (sic) after supper.
ToS
No, I don't believe i'm confusing these two phrases.
Hmm, time & amplitude do exist simultaneaously at any point in a time series of pressures - frequency is a derivative of time. So perhaps the problem is that frequency cannot be determined by a single sample in time of the pressure but only by a series of samples?
Yep
To auditory perception, all that is arriving is a series of nerve pulses from the auditory nerve, not a sinewave - it has no particular mechanism for accessing sinewaves. If in that analysis of the nerve impulses, it finds patterns & these patterns form a regular time interval, then it has the perception of what we call a pure frequency. Again, it has no knowledge of sinewaves, only patterns - regular, random, pseudo-random. Where is the sinewave in the crackling of a fire, the applause , the sound of rain on a roof, etc?
maybe you are thinking of ITD & ILD - how we localise sound sources? Or maybe you are thinking of the different theories of how the inner ear works - place theory Vs temporal theory? i.e. the first is based on frequencies & the second on timings.
Again, I believe you are mixing up things by simply only considering frequencies - auditory processing works to a large extent, using patterns.
Look at the video I posted & the Wiki link above about place theory
No, I don't deny that (the model is consistent but it's just a model) I simply stated that this undue focus on sinewaves is not relevant to auditory perception
Right, Dan, what is meant is the over-exaggerated, ultra-realism that can impress initially but ultimately is fatiguing (talking about audio)
I used it because someone said the ear works like an FFT does - that's all
Yes reality exists - no doubt about that. Our attempts to contact this reality & understand it whether by mathematical algorithms or through our senses are separate & divorced from reality itself - it doesn't need these attempts in order for reality to exist but these attempts are simply that "attempts" - they are not reality itself. And yes, I include perception in that - it is simply an attempted way of interpreting reality which has proven useful for our survival - other animals have other ways of contacting reality
Let's not confuse reality with the various ways of trying to understand it - what my statement should have read was "Again, I'll say trying to shoehorn FFT analysis as the only way to describe auditory processing is a leap of faith (& not very scientific)" it shouldn't have mentioned "reality"
I really don't understand what you are saying here, sorry? I can't see how our early experiences in life could be mainly frequency domain - it just doesn't make sense to me?
Yes.Mountainman Bob said:
Yes, it is real.
You can think of it in that way, but the idea of sampling is not needed. Better to think of a time window within which you have observed the pressure. The longer the time window the more accurately you can determine the frequencies.mmerrill99 said:
Think of an AM signal. If you look at it with a broad bandwidth then you see a carrier wave varying in amplitide up and down. If you look at it with a narrow bandwidth (which implies poor time resolution) then you see the carrier (unchanging) plus some sidebands. Both of these views are equally real and true. To say that one is real and the other is a model is simply to demonstrate lack of understanding of reality and mathematics.
You are confusing two things: a stimulus, and the response to that stimulus. The stimulus is a pressure waveform, or a set of pressure sine waves - the two descriptions are equally real and true. The response is a set of nerve impulses - which can also be described by a (different) time waveform or a (different) set of sine waves.
I thought the original disagreement was over whether music consists of sinewaves? Are you now changing the topic of conversation?
That is probably because like most of us you have not spent much time exposed to frequency domain phenomena, and probably have never developed the physicist/chemist's way of thinking which finds things like frequency (physics) or reciprocal lattices (chemistry) just as real as time or space. On the latter I have some sympathy with you; I always found reciprocal lattices difficult to get my head around yet crystallographers can do this without thinking.mmerrill99 said:
Can music be accurately recreated from these measurements or is it ‘lossy’ for lack of better term?
Is this where extrapolation in digital reproduction comes into play?
Yes, that's what I thought I said? So determining frequency requires a sufficient timeframe to observe the repeating pattern i.e it is a derivative
To someone viewing such plots they are two different plots & have different meanings that is why we often see people stating that the 'grass' at the bottom of the FFT is the noise floor of the DUT.
So perhaps the fault is in the overuse of FFTs without the understanding/qualification necessary to correctly interpret them?
Similar to the overuse of Foobar ABX without understanding/qualifications necessary to correctly use & interpret the results?
No, I don't think I am confusing these things - as already seen here people state that the ear works like an FFT. This view is mistaken, IMO & results from this overemphasis on the frequency domain & sinewaves. What you are trying to say that every sound can be described by Fourier transforms - where are the sinewaves in random noise?
I thought the original disagreement was over whether music consists of sinewaves? Are you now changing the topic of conversation?[/QUOTE]
It depends on the measurement process.
Usally the phase information is dropped and therefore the spectral content can't give sufficient information to recreate/reproduce the original waveform (leaving aside the practical restrictions when measuring in the real world; infinity is easily assessed on paper, but .....)
See for example the plot that was given for the measurement of the 9$ DAC; the spectral content is given (multitone sequence) but as the phase informatione is missing one can't tell how the original signal did look like.
If the phase information is included, reconstruction is possible (basis of sampling) within the real world constraints.
No, it's simply that frequency cannot exist without time so how can someone be exposed to one without the other?
Thanks,
I’m just trying to wrap my head around any use/benefit for Fourier transform in regards to psychoacoustics or perception other than a source identifier.
It surely is fodder for argument! 😛
Last edited:
Scott,
At the moment I have been listening to the early women electronic composers, like Daphne Oram, Laurie Spiegel, and Eliane Radigue. Soon I shall be listening to Pauline Oliveros and Else Marie Pade. All of these women have a decidedly different understanding of musical form through electronics - very different to orthodoxical male bombast. Goodness, I have been hearing and visualising psychic places I never knew existed.
The other male composers you mentioned I have heard of, and maybe I will listen to them another time. I’m sure I will, but right now the women in electronica really fascinate me.
ToS
Sugar and spice and all things nice!ToS
+1 I love Laurie Spiegel as well as Delia Derbyshire, who later formed the band White Noise...I love their album 'An Electric Storm.'
It is tempting to also place Clara Mondshine in this category, and I love the music, but it is actually a pseudonym of Walter Bachauer...great music, though I love this track: YouTube
Howie
Yes, if the USUAL FFTs we see are shown for a chirp sound, it will be the same spectral plot if the chirp is played forward or backward - two very different sounds when listened to.
Hence we are back to the limitations of such FFT representations as usually presented.
Hands up those who believed that those FFTs showed "all there is to be seen" about a sound?
How many people believed that the 'grass' at the bottom of FFTs showed the noise floor of the device being measured?
Now be honest with yourself.
It's a bit like the never spoken "warnings" about doing Foobar ABX tests - seldom have I seen those people who mostly promote them on audio forums issue any warnings/notes about training or approach needed.
Last edited:
- Status
- Not open for further replies.