Yes, and wasn't impressed either.
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
it is.
And phase inversion is like sacrilege then
jsrtheta
Addicted to Fun and Learning
Aczel had the advantage of being right.
As you have politely expressed before.
About what, though?
Absolute phase is inaudible? Upthread is objective test evidence that it is.
Recordings are all in phase? Recording tutorials on phase issues abound and phase reversal is a standard tool of recording hardware and software.
We don't understand it so we should leave it alone? Humans use things we don't fully understand all the time. Plus it's cheap to use.
Last edited:
Raindog123
Major Contributor
Guys,
What does “fft(-x) = -fft(x)” even mean? What is the “x” here - time, frequency, magnitude? Is the “fft()” complex?
I am lost…
In my admittedly limited understanding, FFT is Fast Fourier Transform and x stands for sequence or signal in time or space domain. In our case we are clearly discussing sound amplitude vs. time. FFTs of space domain are often used in guidance system target recognition algorithms. An image of a tank transferred to the frequency domain is quite unique regardless of scale or look angle. Maybe that's how our brains are so good at recognizing stuff...
Fast Fourier transform - Wikipedia
It means that the time domain signal “x” can be flipped over by using the negative of the frequency domain (FFT)... back to the time domain using an inverse FFT.
Usually the FFT is complex so one can consider phase.
And also more often than not… in plotting/graphing, we just look at the absolute maghnitude of the FFT to consider only the power spectral display (Power versus Freq).
Which is sort of a “who cares”, except for maybe the thick glasses wearing DSP engineers?
Most of us would just swap the + and - cables,
But them then, as the saying goes…
Question: “What do engineers use for birth control?”
Answer: “Their personalities.”
Raindog123
Major Contributor
So, what a “-x” stands for then?
I am familiar what a mathematical “[integral] transform” — it transforms a function to another function, with their respective function spaces, usually through its “kernel” K(t,u). I am also familiar with the Fourier transform (whether fast or not
). But I am honestly not sure what @mansr meant by his above statement.
I thought it was the glasses.
Wouldn't the Fourier Transform of a unique function, once moved back into the time domain result in the same function? Or just one of infinite transients that satisfy the frequency content?
I did a project last year that involved a lot of transient synthesis starting from a measured PSD vibration spectrum, to analyze against a highly non-linear system (missile on a launch rail). It is well understood in my field that these transients are not unique. We call them representative and do a mini Monte Carlo to envelope the response.
I didn't want to interject into an EE technical discussion.
Edit: I see now that FTs are complex and retain enough information to recover the original function through the inverse transform. Power Spectral Density does not, so the inverse back to the time domain is not unique. I learned something.
Last edited:
I am sure he will explain when he is able.So, what a “-x” stands for then?
I am familiar what a mathematical “[integral] transform” — it transforms a function to another function, with their respective function spaces, usually through its “kernel” K(t,u). I am also familiar with the Fourier transform (whether fast or not
Whatever the math does, it's supposed to be equivalent to cable swapping - red to black. There is is a Stendhal book by that title - Le Rouge et le Noir. Apropos of nothing...
Last edited:
I thought it was the glasses.
Wouldn't the Fourier Transform of a unique function, once moved back into the time domain result in the same function? Or just one of infinite transients that satisfy the frequency content?
I did a project last year that involved a lot of transient synthesis starting from a measured PSD vibration spectrum, to analyze against a highly non-linear system (missile on a launch rail). It is well understood in my field that these transients are not unique. We call them representative and do a mini Monte Carlo to envelope the response.
I didn't want to inteject into an EE technical discussion.
Well maybe the FFT is more of a math deal?
The FFT is only considered a DSP deal, when one is talking about electrical signals. Or Rabiner’s speech work.
(It started out as a thermal heat flow)
Using FFTs in image processing, tautologically speaking, more in the realm of Image processing than DSP.
So, what a “-x” stands for then?
…
Taking the whole signal and flipping yo the opposite polarity.
i.e. Moving the + cable to - speaker position as is shown on my avatar photo.
…
I am familiar what a mathematical “[integral] transform” — it transforms a function to another function, with their respective function spaces, usually through its “kernel” K(t,u). I am also familiar with the Fourier transform (whether fast or not
Can you link/quote the comment? I am not sure which one you are referring to.
Raindog123
Major Contributor
If x(t) is a function (or signal), then Fourier(b × x) = b × Fourier(x) for any constant b. For b = -1, this corresponds to a polarity inversion in the time domain or a phase rotation of 180° in the frequency domain. It's elementary maths.So, what a “-x” stands for then?
I am familiar what a mathematical “[integral] transform” — it transforms a function to another function, with their respective function spaces, usually through its “kernel” K(t,u). I am also familiar with the Fourier transform (whether fast or not
"Heretics" have the better case:
"Old fashioned bass & treble tone controls and modern "tilt" controls are the answer [to the circle of confusion] and they can be changed at will to compensate for personal taste and excesses or deficiencies in recordings. Sadly, many "high end" products do not have tone controls - dumb. It is assumed that recordings are universally "perfect" - wrong!" -- Floyd Toole
Wrong heading. Scroll up a little to the Linearity section.
Raindog123
Major Contributor
jsrtheta
Addicted to Fun and Learning
Hear, hear. Tone controls should always be available, and bypassable. EQ is extremely important in the recording process. And at home, it has made some music listenable that previously wasn't because of bad recording.
I spent 20 years listening without any EQ at all, per "high end" principles that ignored reality.
Your room is not where the music was recorded. Your equipment is not the equipment used. Your ears are not the ears that recorded the material. You cannot duplicate the original conditions, but you can screw up the sound by failing to EQ.
If EQ is available, and bypassable, why on earth would you deprive yourself of at least having the option? I would certainly have enjoyed a lot more music had I listened to Tom Nousaine's advice back in the day. I have discs that were so poorly recorded that I put them back on the shelf after listening once. Now I don't have to.
The music is supposed to be the point.
Great explantion by Tool. Instead of using my good old fashion treble an bass control i use my DSP slider from Mathaudio room eq to creat my desired custome made target curve easy an fast. An yes because of that after 40+ years suddenly some music i did not play so much (due to partly bad recordings but more bad room acoustics) became listenable.
Similar threads
