________________________________
From: Eric Carmichel <[email protected]>
To: "[email protected]" <[email protected]>
Sent: Thursday, June 27, 2013 4:33 PM
Subject: [Sursound] Of stereo miking, Fourier analysis, and Ambisonics
Many thanks to everyone for your responses and insights (re Giving Precedence
to Ambisonics). I would like to comment on the following two responses:
1. from Jeff
**May I suggest ?Demonstration of Stereo Microphone Techniques,? Performance
Recordings #6 wherein 18 coincident, near-coincident and spaced omni (2 and 3
mic) stereo techniques are compared via a line of loudspeakers mounted at equal
intervals and spanning 10 1/2 feet left-to-right. Each loudspeaker was 2 inches
in diameter and the center to center spacing was 9 inches. An electronically
generated tick was switched to each loudspeaker in turn starting at the center
and moving full right, full left and full right again before ending in the
center. The pros and cons of each technique are unmistakable...
Jeff Silberman**
and
2. from J?rn
**hi jeff,
i think the test you're mentioning is not entirely fair, as much as i like
coincident techniques.
such a setup tests for localisation only, and with wide-band transients it is
quite clear that spaced techniques will lose, and their main advantage (better
perceived spaciousness in stereo-only playback, and better LF response) is not
even considered.
miking is a trade-off. testing individual aspects won't tell us much about
actual musical use.
best,
j?rn**
Eric C. responds
The array of 2-inch speakers is reminiscent of many psychoacoustical
experiments I?ve participated in: More laboratory-like than musical. Note that
the clicks run in a sequence in lieu of random places. Once we perceive a
pattern (e.g. L to R sequence), we begin to fill in the spaces based on
patterns. At least that?s my (intuitive) notion. Also, clicks are among the
easiest sounds to localize. The broadband nature of the clicks provides
multiple localization clues, to include ILD, ITD, and (very importantly) pinna
transfer cues. I have collected data from my personal lab to provide evidence
of this latter claim, and I welcome everyone to review and scrutinize it.
Listening tests were performed using 8 young, normal-hearing persons. A rather
large (2.6 MB) Excel spreadsheet contains all the data. I designed this
spreadsheet to provide descriptive statistics for any combination of listeners
(e.g., group all female participants), stimuli, listening
condition (e.g., unoccluded), or azimuth/location on the fly. You can download
the Excel spreadsheet here (again, it's 2.6 MB):
www.cochlearconcepts.com/stats/hearing_data.xls
Graphical representation of the results (using SPSS) are in the same folder,
and you can see it here:
www.cochlearconcepts.com/stats/Figure_6_96dpi.jpg
As can be seen (and heard!), the broadband stimuli are easy to localize when
compared to tonal stimuli. When participants were donning binaural stereo
electronic earmuffs (net acoustic gain at ear = 0 dB, carefully calibrated to
match earcups), lateralization was accurate. But discerning front-back angles
on same side (L or R; e.g., 60 and 120 degrees) was nearly impossible. This
demonstrates what happens when ILDs and ITDs are preserved, but pinna cues are
lost. You can see spectral and time-domain analysis of the broadband stimuli
here:
www.cochlearconcepts.com/stats/Figure_1_96dpi.jpg
I have to agree with J?rn that the example miking demonstration isn?t all that
fair, and for another reason: How much low-frequency energy can a 2-inch
speaker provide? Although the Fourier decomposition of a transient or *click*
sound may suggest it's a broadband signal, I have reservations about Fast
Fourier Transforms and clicks. My reservation is, in part, rooted in my own
ignorance of math, but I?ll have to state that I don?t believe the ear works
exactly as math would predict. Let me explain...
Fourier series shows that an ideal impulse (Dirac delta function or Kronecker
delta?) can be decomposed into sine waves, but these waves have to begin in the
Paleozoic Era and end when travel to distance galaxies is a reality. There?s
something about the time domain aspect of *real-world* sounds that is missing.
So how does the ear respond?
Imagine the inner ear is comprised of contiguous filters. Perhaps the ears
inner hair cells (IHCs) are akin to the reeds of a resonant reed frequency
meter (which can have very fine frequency discrimination ability). Regardless,
how would such filters respond to click sounds or similar transients? A click
in itself doesn?t last long, and certainly the click's period much shorter than
a fraction of a low-frequency sound?s period. But yet, we rely on IRs and
maximum length sequence (MLS) stimuli for response characteristics as well as
hearing demonstrations. Back to 2-inch speakers...
At least the low mass and confined acoustic centers of tiny speakers make them
ideal for transient bursts, but then what instruments aside from a woodblock
does this represent? There always seems to be trade-offs when it comes to
miking, and some of these have to do with mixing down to mono for non-stereo
broadcasts.
There are two additional Excel files in the aforementions stats folder. One is
titled keyboard:
www.cochlearconcepts.com/stats/keyboard.xls
This file merely generates piano keyboard note frequencies for any arbitrary
A4. Throughout history, I believe A4 has changed (e.g. A4 = 376 Hz in lieu of
440 Hz). I had developed this spreadsheet because of a discussion that goes way
back. I am not a musician, but am learning music theory from books and CDs.
The last Excel file has to do with Ambisonics: It was my speaker positioning
calculator based on the corner of a room being the origin of a Cartesian
coordinate plot. Units are English because my tape measure is in feet and
inches. Instead of using a compass, radians, and the center of a circle, I used
basic (x,y) coordinates relative to walls to accurately position loudspeakers.
Probably not worth explaining further, as this calculator is somewhat specific
to my listening room. But if anybody can use it, please do so.
Kind regards,
Eric
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<https://mail.music.vt.edu/mailman/private/sursound/attachments/20130627/0c874042/attachment.html>
_______________________________________________
Sursound mailing list
[email protected]
https://mail.music.vt.edu/mailman/listinfo/sursound
