The main question I will be trying to answer with this lab is the following: At what point does the length of the ramping up and down of a loud noise on top of another softer noise get perceptually segregated into two different sources?
by Ben Radovitzky
In order to find this cutoff, I will be changing the amount of time it takes for the louder noise to ramp up and down. I will be decreasing the length of the ramp time from about .25 seconds to 0 seconds (instantaneous jump). As we already know, the slow ramp time should appear to sound like one oscillating sound source, and the instant one should sound like two sounds, one consistent low volume one, with a short repeating tone played on top of it. The whole sound clip will be repeated several times in each case so that the effect is easier to notice.
If all goes well, it should sound like the noise passes from being a single oscillating source to sounding like two distinct sources at some specific point. It is possible that there will not be one exact point for everybody, and that it depends on the person. In this case, I will hope to do more testing and find an average point at which this happens. With more time, I would have liked to run similar tests with different frequency tones and different sound types as well, including speech and other natural or commonly heard sounds.
Here is the sound byte. I couldn't figure out how to add my voice to the beginning, so here is the explanation for what you are about to hear. A flat gaussian noise will be played consistently for 18 seconds. There are a total of 6 different "B" noises played over the base noise, each of which is repeated 3 times in a row. This is done to get a bit accustomed to the noise, and to be able to better figure out whether it sounds like a single noise or two. The ramp times will get shorter and shorter as the sound clip continues. At some point, it should sound like the noise passes from being one oscillating sound source to being two sources, one quiet noise and one shorter repeating louder noise.
Bregman, Al. "Audio Demonstrations of Auditory Scene Analysis - Homophonic Continuity and Rise Time." Al Bregman's Website, 2008, webpages.mcgill.ca/staff/Group2/abregm1/web/downloadstoc.htm#32.
Comments
Emily Huang
a) For me, the noise sounded like a single source until about 13 seconds in, after which it sounded like a puff over constant noise and then a sharper sound over constant noise as the ramp got even shorter.
b) This demo definitely helped answer your question! Obviously, everyone's boundary will be subjective, but I think the demo does a good job of helping a participant figure out their boundary.
0
Ben
@Emily Huang,
c) Not only does this answer my question perfectly, it also follows my hypothesis and shows that different people have a similar cutoff for this kind of segregation.
d) I think the main issue with my illusion right now is that I don't show or tell the person taking the test how long the ramp times are, and that I have a very coarse idea of where people stand. If I had a chance to do this again, I would certainly make the "start" and "end" ramp times a lot closer together, and make the differences between them smaller and less noticeable, to find a more precise average where people tend to start the perceptual segregation.
0
Griffin Leonard
a) The clip sounded like a single oscillating sound source until around 14 or 15 seconds into the clip, where it began to sound like two separate sound sources (one that sounded like constant noise and a another sharper, louder sound as you described above).
b) This illusion answers the question. It demonstrates that as you increase the slope of the ramp into the louder sound, there is a point where the sound is perceptually segregated into two sounds (for me it is whatever the ramp's slope is at ~14 seconds into the clip).
0
Ben
@Griffin Leonard,
c) Like with Emily, it seems like you had very similar results that answered my question and approached my hypothesis.
d) Since I already discussed my shortcomings on this experiment, I will use this space to ask my follow-up question: How does this effect change when we vary the types of sounds we are hearing? In other words, would we get different results if the sound was a tone, voices, or other sounds that aren't just gaussian noise? Do the ramp times vary based on type of source?
