To get a good frequency accuracy with an FFT (or more generally a DFT), you need to have a small bin width. The bin width is inversely proportional to the time interval over which you look, so you need a long time interval. When you calculate a moving DFT over long time intervals, you don't know accurately when the tone started or ended.
The Hilbert-Huang transform is meant to solve that issue. It's an algorithm to split a signal into a bunch of amplitude- and frequency-modulated sines.
The Hilbert-Huang transform is meant to solve that issue. It's an algorithm to split a signal into a bunch of amplitude- and frequency-modulated sines.
Maybe one way to identify musical notes. However, the Hilbert-Huang transform was first patented by NASA in 2001. OTOH, Autotune was introduced in 1997. Could be that Melodyne uses Hilbert-Huang transform for some of its (IIRC unique) capabilities. https://www.celemony.com/en/service1/about-celemony/technologies
According to https://en.m.wikipedia.org/wiki/Auto-Tune Auto-Tune doesn't use anything like the Hilbert-Huang transform, but a different method based on autocorrelations.
Maybe the 2008 plug-in that you linked to, the one that identifies notes in polyphonic music, uses it, maybe not.
In any case, I haven't seen any claims of uniqueness in the book so far, so it could be that someone else invented something similar.
This discussion is very off topic. Do we stop it or ask the moderating team to put it in a separate thread?
Maybe the 2008 plug-in that you linked to, the one that identifies notes in polyphonic music, uses it, maybe not.
In any case, I haven't seen any claims of uniqueness in the book so far, so it could be that someone else invented something similar.
This discussion is very off topic. Do we stop it or ask the moderating team to put it in a separate thread?
Last edited:
We were talking about phase, maybe of LF transients (say, for example, a kick drum, maybe with a plastic beater to add a little 'click' to the sound), and possibly even measurements thereof.
Typical audio FFTs are steady-state and phase-free (magnitude only). Can't measure time-domain transients that way.
Why don't we have a better way to measure time-domain transient performance?
If we did it might also help measure dynamic imaging differences between a stereo amp and a dual mono-block setup?
Can we make effective us of more modern signal analysis techniques, maybe something like Hilbert-Huang?
Typical audio FFTs are steady-state and phase-free (magnitude only). Can't measure time-domain transients that way.
Why don't we have a better way to measure time-domain transient performance?
If we did it might also help measure dynamic imaging differences between a stereo amp and a dual mono-block setup?
Can we make effective us of more modern signal analysis techniques, maybe something like Hilbert-Huang?
When talking about polarity we have to face that all positive waves are gone through npn transistors and all negative waves through pnp types. Thats a fact in nearly every amplifier and in most preamps even in D/A converters. So perhaps this is a further reason we can hear these differences when changing polarity.
It would be interesting to compare the polarity with a sound file reorded with a class A microphone direct to an A/D converter and then played through a single ended class A amp.
It would be interesting to compare the polarity with a sound file reorded with a class A microphone direct to an A/D converter and then played through a single ended class A amp.
An example of absolute phase might be if you mic'ed a kick drum at the front head. When struck by the drummer at the rear side, the front head moves out towards the audience. So, maybe a reproduction speaker should do the same?
This has been a fascinating discussion. However, since most speakers don't pay attention to phase coherence or time alignment in their design, much of it may not apply to home listening. Not to mention that we are completely at the mercy of decisions made in the recording studio.
I have preferred listening to my Vandersteen 2C speakers for over 20 years. It is hard to find others that do as well at preserving the impact of percussion instruments. The Theil designs are similar in their dedication to preserving both phase coherence and time alignment. Big planar magnetic speakers and electrostatic speakers also do a good job, but it can be hard to find ones that extend well into lower bass frequencies.
The Pass designed Slot-loaded Open-baffle (SLOB) speakers that a few of us got to build last month have an advantage in their use of a single full range driver that covers most of the audio band. The big 15" drivers that cover the lower bass frequencies. They just need amplifiers with the same phase for bi-amping (or taking the phase into account with the speaker connections). Mine are still going through the run-in and level adjustment process. They sound very good so far.
I have preferred listening to my Vandersteen 2C speakers for over 20 years. It is hard to find others that do as well at preserving the impact of percussion instruments. The Theil designs are similar in their dedication to preserving both phase coherence and time alignment. Big planar magnetic speakers and electrostatic speakers also do a good job, but it can be hard to find ones that extend well into lower bass frequencies.
The Pass designed Slot-loaded Open-baffle (SLOB) speakers that a few of us got to build last month have an advantage in their use of a single full range driver that covers most of the audio band. The big 15" drivers that cover the lower bass frequencies. They just need amplifiers with the same phase for bi-amping (or taking the phase into account with the speaker connections). Mine are still going through the run-in and level adjustment process. They sound very good so far.
My speakers are time correct, but the latter is indeed a major issue leaving much of the sw with mixed up phase.
dave
A minimum phase EQ curve only fully restores phase response if the applied curve is an exact mirror of the minimal phase curve that was applied first. This is not a situation that occurs in the recording studio.
I like to do unconventional things when I have the time during a recording session, just to hear how it sounds. That's how I discovered that the Neumann Gefell M582 with an M94 capsule is a great close mic for cabs. It just sounds right. It is my standard now (often with a close ribbon and a distant LDC).
Might be worth trying out!
Last edited:
Say oooh into a microphone and monitor the signal throughout the signal chain to see if it is upside down.
If there is a difference i agree that it wil be extremly subtle. Maybe you have a link to that paper?
It's the article of Greiner and Melton, "Observations on the audibility of acoustic polarity", Journal of the Audio Engineering Society, April 1994.