Wasn't psychoacoustics temporal/phase sensitivity research measured with fairly simple waveforms composed of a small number of sinusoids? Real music is much more complex. Perhaps consider the sound of a stick striking a cymbal. Its not a particularly LF instrument. There is a ping sound on the attack that is dependent on how and where the cymbal is struck, the type stick material, how the the stick is manipulated by the percussionist, etc. On many systems most or all of that transient detail is lost. The point remains that HF transient information of that complexity and frequency content can be audible in a live situation.
In addition IIRC at one time @1audio pointed out that while humans are insensitive to fixed phase differences at higher frequencies, changing phase is much more audible. Phase of source material can be changing because signals such as speech and music signals are non-time invariant and nonstationary. So are all physical systems for that matter. It's just a question of to what extent.
An interesting quote below from a published book review of, "Hilbert-Huang Transform and Its Applications:"
"The Hilbert-Huang Transform (HHT) represents a desperate attempt to break the suffocating hold on the field of data analysis by the twin assumptions of linearity and stationarity. Unlike spectrograms, wavelet analysis, or the Wigner-Ville Distribution, HHT is truly a time-frequency analysis, but it does not require an a priori functional basis and, therefore, the convolution computation of frequency. The method provides a magnifying glass to examine the data, and also offers a different view of data from nonlinear processes, with the results no longer shackled by spurious harmonics — the artifacts of imposing a linearity property on a nonlinear system or of limiting by the uncertainty principle, and a consequence of Fourier transform pairs in data analysis. This is the first HHT book containing papers covering a wide variety of interests. The chapters are divided into mathematical aspects and applications, with the applications further grouped into geophysics, structural safety and visualization."
https://www.worldscientific.com/worldscibooks/10.1142/5862#t=aboutBook
In addition IIRC at one time @1audio pointed out that while humans are insensitive to fixed phase differences at higher frequencies, changing phase is much more audible. Phase of source material can be changing because signals such as speech and music signals are non-time invariant and nonstationary. So are all physical systems for that matter. It's just a question of to what extent.
An interesting quote below from a published book review of, "Hilbert-Huang Transform and Its Applications:"
"The Hilbert-Huang Transform (HHT) represents a desperate attempt to break the suffocating hold on the field of data analysis by the twin assumptions of linearity and stationarity. Unlike spectrograms, wavelet analysis, or the Wigner-Ville Distribution, HHT is truly a time-frequency analysis, but it does not require an a priori functional basis and, therefore, the convolution computation of frequency. The method provides a magnifying glass to examine the data, and also offers a different view of data from nonlinear processes, with the results no longer shackled by spurious harmonics — the artifacts of imposing a linearity property on a nonlinear system or of limiting by the uncertainty principle, and a consequence of Fourier transform pairs in data analysis. This is the first HHT book containing papers covering a wide variety of interests. The chapters are divided into mathematical aspects and applications, with the applications further grouped into geophysics, structural safety and visualization."
https://www.worldscientific.com/worldscibooks/10.1142/5862#t=aboutBook
Last edited:
At least for localization the transition happens at head dimension wavelength. Above which its mostly spectral as high frequencies are shadowed by the head and below sound goes around the head and its more about time/phase and a transition region in between. Search for ITD and ILD to get more info. There is probably some other aspect to perceiving sound than this localization mechanism but this is quite easy one to remember and only one I remember so perhaps a start if you go look after the stuff 🙂
Recording to tape was a completely different technique than is used today. It depends at what industry level (budget) your talking about.
In general though the whole band would be set up with amps & drums in as much isolation as possible and micing and desk setup as close to final mix as possible then a lot of the time bands were recorded live with lots of compression and overdubs were only done for mistakes, extra vox etc. Musicians were used to that in those days. Post production was a nightmare at times compared to digital editing these days ... everything was more of a performance ... 🙂
Large format tape though is magical ... everything you record on it including mistakes sounds better when played back ... the exact opposite of digital ...
I was thinking there may be a similar transition region for polarity sensitivity.
^^ Same analog recording methodology can be used with digital recordings as well, playing full takes with a band, but perhaps the musicianship on average is worse as everyone is now an recording engineer, musician and producer all the same time and more often there is need to go fixing stuff, or do one track at a time as there is only one or few musicians involved. Its magical when a full band makes a good take, hard to better one track at a time. Best place to enjoy high fidelity music is at live shows and studios, live, before too much prosessing and medium gets involved and reduce the magic.
The Philips studios (now Polyhemnia) in Baarn used ESL-63s at least for classical recordings.
Wrt the audibility, it depends (on a lot), basically, the human hearing sense is sensitive to polarity.
Humans can accurately hear frequency changes up to about 4 kHz to 5 kHz, far less accurately at higher frequencies. It is usually explained with neural firing patterns locking to the waveforms up to 4 kHz...5 kHz. No idea what this means for polarity audibility.
Yeah a good live take is the same (performance wise) but the sound on playback is NOT ... it's nothing to do with methodology it's analog tape and circuitry that made the difference. The ONLY good thing about digital recording is the speed of editing ... sound wise there is no comparison (even with the avalanche of noise adding plugins) ... 😱
IBP works over a span of freq but it is limited. If you have access to one hardware ( or plug in) IBP you can compare between allpass and time aligning the signals on takes needing compensations , there will be difference in outcome.
Is it a deal breaker? Not in my view as IBP can be a great help and is fast to implement ( time aligning can be time consuming and during a session you most always fight the clock).
About mixing analog and digital: nothing stop anyone to track in analog and then digitalize at later stage... or the other way around! I've done both to great effect.
You don't know in advance how a track will sound in a mix BUT you can know if the multi mic induce some 'phasing': for this we have tools to visualize ( lissajous figure) and the most lmportant button on monitoring section called 'mono'. When locating mic seting up amps i monitor in mono and compare mics to the 'main one' ( usually a sm57) modifying location until it sound good and 'full'. Doing so will limit the amount of work at later stage ( anyway and beside what people usually think you can't do many things in 'post production' stage if it doesn't sound good at takes and better spend time wisely, 1hour to set up save 3 to 4 of editing...).
That said I most always now what i want - or what musician want- (kind of sound) and which mic ( or blend) to use to reach it ( tracking cab is for me almost always: sm57/e609/md421/ ribbon ( Coles, Aea, Beyer M160.../ large capsule capacitor -u87/47 fet/67/c414/...).
This is my typical set up. Most of the time max 3 mic open, more probably 2. Large condenser located away from cab to catch early reflection, the 4 drivers at once and room/air ( used most only on soli the 47fet being the exception - works well for rock/blues in 'close' micing) ( typical cab 4x12").
But this is a way to do it, there is many more and possibly better ways to do it. In no way i claim to have THE answer or THE workflow.
Last edited:
Of Sound Mind: How Our Brain Constructs a Meaningful Sonic World https://www.amazon.com/gp/product/B08RHGZBT2/ref=ppx_yo_dt_b_d_asin_image_o05?ie=UTF8&psc=1
Last edited by a moderator:
Not necessarily, they also use clicks, bursts and filtered noise a lot.
Of course.
Still, when I say "book" into a microphone, I don't see the "oo" waveform's polarity change.
Non-stationary is clear, but what does time invariant mean for signals?
I only know the term for systems with one or more inputs and one or more outputs; if shifting the input signals in time always causes an equally time shifted but otherwise the same output signal, then the system is time invariant. It's unclear to me how to extend that from systems to signals, as signals have no input.
It's clear that musicians can be regarded as time variant systems, though. When asked to perform (input signal), they perform better on some days than on others (output signal). They are also non-linear: when you ask them twice as loud to perform, they won't necessarily play twice as loud.
The reason for using linear time-invariant approximations is to simplify calculations: change differential equations into algebraic equations in the Laplace or the z-domain, and being able to specify a system's behaviour without knowing the input signal. I wonder how much of that simplicity they can retain without those approximations.
If you mean the "oo" sound is not perceived as a time-domain transient, I would agree.
There is a difference between physical signals on the one hand, as opposed to signals as mathematical or conceptual abstractions on the other hand. I was referring to physical signals. For example air pressure impinging upon the ear (or a upon a mic diaphragm) as function of time when a piece of music is played back will never be perfectly the same each time playback is repeated. Random noise affecting physical signals could be one reason.
Would agree completely that simplicity can be very useful as a practical matter. Simplicity is one reason for measuring with steady-state and or phase-free FFTs. The problem with that is WYSIATI. There are a lot of people who don't keep in mind or who never understood, maybe were never taught, that we do some things for simplicity sake, not because the simplified view is the whole story. https://www.google.com/search?q=wys...1i433j0i512.6979j0j1&sourceid=chrome&ie=UTF-8
Depends if your recording isolated tracks or a live room.
Between recording microphone locations,
the room, the mixdown, playback amplifier, playback EQ, playback speakers, crossovers etc etc
absolute phase lol....nope.
Far as recording drums and hearing different echo slapback , room sound.
similar when trying to capture room sound with a piano.
Depends, using old Tri mike, one kick drum mike and 2 overheads.
usually its bleed from the kick.
you can invert the channels on the overheads, or invert the kick mike.
either way the room or intensity of the echo sounds different.
depends on the pickup pattern of the overheads as well.
I dont think anybody really over thought absolute phase etc etc.
you just put a little tent over the kick drum and mike with a blanket.
then inverting the channels didnt do so much.
just isolated the kick, or put the mike closer to the head.
but its a little to much bong or skin ring if you mic close.
heck in the 80's 90's depending on genre or at least working bands in clubs.
cut a hole in the front head, shoved the darn mike into the kick drum.
or in the studio the front head came off. and used a blanket
Between recording microphone locations,
the room, the mixdown, playback amplifier, playback EQ, playback speakers, crossovers etc etc
absolute phase lol....nope.
Far as recording drums and hearing different echo slapback , room sound.
similar when trying to capture room sound with a piano.
Depends, using old Tri mike, one kick drum mike and 2 overheads.
usually its bleed from the kick.
you can invert the channels on the overheads, or invert the kick mike.
either way the room or intensity of the echo sounds different.
depends on the pickup pattern of the overheads as well.
I dont think anybody really over thought absolute phase etc etc.
you just put a little tent over the kick drum and mike with a blanket.
then inverting the channels didnt do so much.
just isolated the kick, or put the mike closer to the head.
but its a little to much bong or skin ring if you mic close.
heck in the 80's 90's depending on genre or at least working bands in clubs.
cut a hole in the front head, shoved the darn mike into the kick drum.
or in the studio the front head came off. and used a blanket
I think around 400 Hz
If I remember it is related to skull size, and distance between the ears.
But there is additional brain function, which compensates for our physical ear positions/ skull size.
Without much detail it what makes binaural beats possible
and few other acoustic tricks possible.
Because of the additional brain processes for sound.
I guess electrical equivalent be 2 microphones is our ears.
But the brain applies additional DSP to the signals LOL.
assumed for protection from predators, and being able
to give more directionality to low frequency. Aka footsteps
or approaching prey
It's off topic, but one thing that still amazes me is that linear time-invariant models augmented with a rule of thumb can be used for designing high-order single-bit sigma-delta modulators. The quantizer is as non-linear as it gets, but by pretending it is an amplifier with some arbitrary positive gain and additive noise, you can actually calculate usable loop filter coefficients, provided you don't make the out-of-band noise gain larger than about 1.5. It makes no sense at all, but it works.
It's an almost exact quote from the preface of the book, which I borrowed from the Delft University library. I guess the Hilbert-Huang transform might be an interesting method for automatic recognition of musical notes.