Greetings to All, As always, many thanks for everyone’s time. This post references replies received from Peter and Jörn: Thank you both for your expertise, sharing experiences, and taking time to write.
[from Dr. Lennox] “For mobile listeners, and indeed, off-centre listeners, the amplitude gradient at Lf was readily discernible, so that kind of cue, though not normally part of Duplex theory for directional hearing, seemed relevant to the experience.” ELC: I recently made recordings in a semi-anechoic room using a third-octave stepped-frequency test signal. I presented the tones at different angles.* The lowest test frequency was 50 Hz. At this frequency, at least in the semi-anechoic room, sound-source location was unambiguous. I wasn’t testing my ability to localize; this was just a casual observation that seemed to go against the grain. Once I process the recordings, I’ll have a chance to hear and see how well defined source-source location is as a function of frequency (all I have now are the A-format files recorded to the TASCAM D680 via my TetraMic). It'll be curious to hear whether the 50 Hz tone appears to come from a specific direction in my horizontal array (which consists of 8 full-range speakers identical to the speaker I used in the test room). *Notes regarding test setup: I did not use a turntable for rotation (the speaker was fixed); instead, I have a ball head mounted on a tripod. A separate jig allows me to center the mic vertically over the ball head. I rotated the mic in fifteen degree steps, 0 to 360 degrees. The stepped test tone was presented and recorded at all 24 positions. [from Jörn] “...the inability to localise bass sounds is a very persistant urban myth. In rooms, ok, but anybody who has been near an open-air rocknroll stage during subwoofer calibration will have no trouble localising the sound :)” ELC: Visual cues may play part in this ability, too, but I fully agree with you: The sound-source direction is unambiguous in certain situations. This is why I wish to include multiple subs in my forthcoming experiments. Another common myth is how much the Doppler Effect plays in our perception of an object moving towards or away from us. A car moving towards us will have a constant, albeit upwardly shifted, frequency. There is no frequency change until the moment the car passes. Note that I selectively use the word frequency, not pitch. Perceived pitch DOES change, but is a function of increasing intensity, not a constantly-changing frequency. Pitch, measured in mels, is affected by SPL. Until the moment a moving object passes the observer, it’s mostly changes in intensity and quality (timbre) that alert us to movement. I’ve used the Doppler Effect functions in various VSTs (or RTAS for Pro Tools), but mostly to provide the effect of an object moving on a curved path. I bring this up because of a subsequent comment (below) regarding rendering. [from Jörn] “...Interesting research project! If you can, please share the results.” ELC: I’ll be glad to. Not every listener is human, hence the need to remove assumptions based on human hearing. Listening “machines” are involved (more later), and arrays of mics along with beam-forming techniques can determine both time and level differences over a wide range of frequencies. Furthermore, the mic arrays are considerably more directional at the low frequencies than, say, the typical (first-order) cardiod mic. Without a barrier (Jecklin disc or head/torso), no ILD. Mics close together = skewed or non-existent ITD. Just want the waves in centre of array to be as they would exist in real life... in the absence of a human head. [from Jörn] For the actual experiment, I can see how you would use pre-rendered files for extra robustness, repeatability and foolproof documentation, but while finding the setup, I'd really recommend to use real-time filtering. Insert usual ‘linux/jack/ardour/fons' stuff’ plug here.” ELC: Yes, thanks for suggestion. Because I won’t be only person running the experiment, I will be making the files playable on the most generic of DAWs. It’s a bit like submitting audio stems to a video producer: I don’t always want the receiving party to have control of all the effects, just the overall level or balance. Secondly, I may use filter functions that aren’t available in my VST collection. Admittedly, I can’t at this time imagine needing a filter that requires advanced signal processing (e.g. MATLAB or LabVIEW) offline. [from Jörn] “...The thing with dual stacked rings is: sources on the equator have very low rE already, being ‘vertical phantom sources’. Nothing against rings as such, but make sure you have a good part of your speakers at ear level. It sure is a psychoacoustic decision, but if it really bothers you, you should go for an even distribution on the sphere anyways.” ELC: The general consensus is to use an even distribution on a sphere. The original reason for considering three vertically-aligned hexagonal arrays is: 1) providing “unobtrusive” space for video (the logistics of the video and the visual stimuli are still in the making); 2) not able to fly heavy or large loudspeakers from ceiling; 3) making smaller speakers symmetrical (above and below) subs as well as listening position. RE number 3: I believed that time alignment for complex sounds whose constituent frequencies will emanate from both subs and satellites would be more phase cohesive in this arrangement than simply putting an array of subs on an arbitrarily-chosen plane located towards the floor. I'll be the first to admit that my notions are often wrong or (less significant) totally overkill. I have no problem with either overkill or being proven wrong. [from Jörn] “...but it's quite easy to render, say, a passing car, together with its characteristic floor reflection in case that’s relevant for cars, in higher order... getting a complete ambient recording in HOA is hard and involves something like the eigenmike, but ‘artificial’ stimuli can just be panned, exploiting the improved focus of HOA. ELC: Yes, sound design for a few vehicles moving simultaneously can be done without much trouble. For a general audience, L-R panning works well, particularly when a large part of their concentration is on the video. Adding lot of moving objects is more difficult because they will have overlapping interaction--the degrees of freedom grows exponentially. I have to be careful that I don't substitute physical wave propagation accuracy for perceptual realism. Although we can only give attention to two or three moving sources at a time [any references as to an actual number, I don’t think it’s large], a highly complex scene may better be left to an Ambisonic recording mic that can pick up every nuance in one fell swoop. A HOA mic would be wonderful, but can't afford one at this time. [Jörn, re 400/700 Hz crossover frequency] “...not if you wish to employ conventional woofers.” ELC: Agreed, but rare is the day I do everything in a conventional manner. I suppose “thinking outside of the box” has dual meaning when it comes to loudspeakers :) Again, many thanks for the ideas, help, and encouragement. Eric C. (a.k.a. ELC) -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://mail.music.vt.edu/mailman/private/sursound/attachments/20130224/147f1099/attachment.html> _______________________________________________ Sursound mailing list Sursound@music.vt.edu https://mail.music.vt.edu/mailman/listinfo/sursound
