Wednesday, November 26, 2008
After listening to this sound for a while I realized that I was actually using it quite effectively to echolocate. Possibly more effectively than clicking. And I can see where the myth of "pressure on the face" comes from. Since this sound was more omnipresent than the sporatic clicking you get a much better transition between the "presence" of different objects.
This makes me wonder, if some white noise were to be emitted at, say, the belt buckle level (somewhere away from the ears, but a source that travels with you) if that would be of great benefit for echolocation. Maybe clicking is not the way to go?
Anyone out there ever notice this phenomenon? Let me know what you think.
Saturday, November 8, 2008
After doing this for a while you'll start to notice the slight differences in the sound, and you'll be able to make little tweaks in order to improve it. (See article: "Properties of a good Click") Even if you don't think you're noticing, you may very well be subconsciously learning, as a great deal of echolocation is achieved by subconscious recognition. You can change the frequency by changing the shape of your mouth, and if you continually make the sound you will just get accustomed to it, and gradually it will become easier to use it for echolocation.
Tuesday, November 4, 2008
The "Cluck": Made by lightly pressing the tip of the tongue against the roof of the mouth and then breaking the vacuum and smacking your tongue against the floor of your mouth.
The "Giddyup": This one is made by breaking the vacuum and drawing air in between the sides of the tongue and the molars, and is commonly used to communicate with horses. This signal, by nature, is produced at the sides of the mouth, and therefore is emitted away from the sides of the head. This is interesting in that we can send the signal to either one side or the other, but it is more difficult to send the signal directly to the front with this method.
The "Blade Pop": This one is the most difficult for me, but sounds like the one used by many proficient echolocators. This requires that you suck the blade of your tongue (the big meaty part in the middle) up against the roof of your mouth until you've got a good amount of surface-to-surface contact, and then break the vacuum by pulling your tongue away. This one requires significantly more vacuum than the previous two. When executed correctly, it sounds distinctly like a bottle cap being depressed or released.
The frequency of all three clicks described above can be adjusted slightly by the shape of the mouth. Generally a wide mouth or smile will generate a higher primary frequency, as well as make you appear happy :)
Below, I've uploaded the waveform generated by each of these clicks, as well as a spectrum analysis. The waveform shows the amplitude of the sound over a certain time period (as indicated), and the spectrum analysis is a plot that lines up with the waveform and shows the distribution of frequencies that occur within the sound signal. Brighter colors meaning higher concentration of waves in that region. Higher frequencies are at the top and lower frequencies at the bottom of the spectrum analysis.
The Cluck (200ms):
The cluck waveform is neat because you can distinctly see two spikes. The first small spike is the tip of the tongue separating from the roof of the mouth, and the second spike is the tongue smacking against the bottom of the mouth. Although the latter is significantly more prominent than the former these two sounds are within 10-15 ms of one another and have the potential to cause interference to the listener. This signal could introduce ambiguity.
The spectrum analysis shows that most frequencies coming from the cluck signal are quite low. The higher the frequency, the more energy it has and it also allows for better recognition and better resolution. (See Properties of a Good Echolocation Click) A broad distribution of frequencies would theoretically give you a reliable signal since some frequencies will be absorbed by objects and others will be reflected depending on the resonant frequency of the material or object.
The spec to the right of the actual signal is not an echo, but actually a drop of saliva swishing around in my mouth as an artifact of the signal creation.
This is the signal I started with, but have not had a lot of luck using it effectively. I think this is primarily because of the concentrated low frequency and "double pop".
The Blade Pop (100ms):
Notice the sound envelope as compared to the "Giddyup" below. The "attack" (time it take for the signal to get from zero to peak amplitude) is much less for this signal. Approx. 7ms for the Giddyup, as opposed to about 1ms for the Blade Pop. This gives the signal more of a distinct "pop" which will actually impact objects better and thus be reflected better. Think of it has a "harder" signal. A superball bounces better than a sponge. In other words, the object can be absorbative making it difficult to bounce a signal off of, but the signal can also be "squishy", making it easier for the object to dampen the impact as opposed to reflecting it.
Notice that there is fairly good distribution of the signal although there is a blank spot, and there could be more in the high end which would make for a better signal. If you look at the spot about 70% through the spectrogram, you'll see a faint echo of the higher frequencies. This is approximately 70ms from the generated signal and probably corresponds to the shape of the signal bounce pattern in the room. A 70ms delay would mean approximately 70 feet of signal travel, so this signal is more like an echo after the signal has bounced around the room a while as opposed to the instant ricochet off of the nearest wall.
This being the sound that I am struggling with, it may be that, with practice, I could distribute the signal over a wider range of frequencies and become more accustomed to its sound.
The Giddyup (300ms):
The signal itself has a 7ms attack as discussed above, and the signal itself is about 25ms as opposed to 12ms for the Blade Pop.
Good distribution of the signal, a little ricochet at about 70ms on the higher frequencies. As mentioned above, this one is directed away from the sides of the face, which may or may not be a good thing. It's nice that it is more inline with the ears, but then you have to turn your head slightly in order to notice objects directly in front of you.
This is the signal that currently gives me the most accuracy. I am accurate within 1/2 inch or so of flat walls whereas the Blade Pop only gives me accuracy down to 6 inches or a foot. I will need to play with that signal a bit more and report back, because I do like the distiction of it. It is much more "poppy" than the Giddyup.
I will need a more sound proof room and more controlled environment in order to directly observe the reflexive properties of certain objects with these clicks. It's a good thing the human brain is faster, smarter and more acute than any computer, otherwise we'd still need our eyeballs to see things.
If I try out different clicks and postulate which click I think works best that would be extremely subjective and I’m sure that different clicks work better for different practitioners. However, there are many fundamental qualities of a click signal that make the signal better suited for echolocation:
Signal Frequency. The frequency of a signal governs the resolution, in that a shorter wavelength (higher frequency) will give you more definition as to what it has bounced off of. Low frequency waves, since they have a longer wavelength are not as distinct. It has been said that the region of 3kHz is a good place to be for echolocating.
It is my hypothesis that a broad distribution of frequencies would be desirable (above a certain value) so that if some frequencies are absorbed by an object, others will be reflected by it.
Signal volume. The sound must be loud enough to stand out over ambient noise. 40 dB is about the level of quiet speech from a few feet away.
Clarity. This is probably one of the most, if not the most important property of the sound. It is critical that after the sound is made, there are no artifacts of the signal source still emitting sound. In other words, the sound must stop abruptly so that the reverberations can be clearly heard. If the sound were to taper off at all, this small amount of sound would easily cover up the reverberations, or at least create a confusing blend of signals.
Directional. If the signal is omni-directional, IE, if it is the same volume in all directions from the source, it will be difficult to know the direction from which it is being reflected, and thus where the object is that is doing the reflecting. Think of it like a flashlight. You can point a flashlight at one object and see what it is. Whereas, if you switch on a light bulb you are able to see everything in the room and there is more information to take in. When echolocating, we want to eliminate as much excess information as possible.