Monday, October 19, 2009

Fun with Audio

More numbers than the mind can comfortable contain...

Back at home, I discovered that the video camera did indeed capture at least one double knock. Once I got the video loaded up on my computer, I set about seeing what I could extract from it. You may want to have the video clip open and handy for reference. Also good for reference is this map of our locations at the time (click for a larger version):

The three double knocks flagged on the video clip I posted in my last entry are the ones I am sure are there, and are the ones I heard in person. They occur at the following times (all central daylight):


They are at very even 10 second intervals, exactly as would be expected if they were the simulations. However, if Scott had started the simulations at or within a few seconds of 7:00:00, as he believes he did, and did only 5 double knocks, as he believes he did and actually did do every time I definitely heard him, then his last simulated double knock should have been at 7:00:40. Allowing about 5 seconds for the sound to propagate the mile to my location, his final simulated double knock should have reached my camera at about 7:00:45. But actually this is when the first one showed up, not the last one. The final double knock on the tape is about 20 seconds after Scott should have finished.

Did Scott start 20 seconds late? He is pretty certain he did not. But, even if he did, where are the first two simulations? At 10 and 20 seconds before the first marked double knock at 7:00:49, there are no comparable loud crisp double knocks. As the camera was pointed most directly towards the double knocks at the beginning and ending of the tape, you would expect those first two simulations to have been as loud as the last one unless something changed dramatically in that very short interval of time.

Everyone who has reviewed the tape agrees that to the ear these are very good double knocks, with quality and cadence that are a close match to the real thing. Those who have examined them include some of the PIs from the Cornell project and quite a few people who are familiar first-hand with the double knocks of tropical Campephilus species. Only the final double knock is loud enough to render clearly on a sonogram:

Double knock #3:
This is a perfect Campephilus double knock signature: two crisp sharp raps, heavy in the base tones, about 70ms spacing between the two, first knock distinctly louder than the second but both quite similar in quality. One of the things that I note about it both on the sonogram and directly by ear is the crispness of the raps. There is not a trace of "smearing" from echo or dispersion. Granted, it is not very loud, but it is loud enough to image clearly on the sonogram. It remains very hard for me to imagine how these sounds could have traveled through 1800m of forest and yet retained such sharpness. Sound waves just don't do that.

The sonograms for the other two double knocks are fainter and harder to make much of. In all these sonograms, there is a steady whine at about 500Hz; annoyingly this is just above the dominant frequency of the double knocks. This is an artifact created by the camera motor:

Double knock #1:
Double knock #2:
The video camera was pointed away from the source in these cases. I am surprised that #2 appears the faintest on the sonogram; to my ears it sounds louder, or at least more distinct than #1. It is straddling the vertical grid line at the center of the image, if you are having trouble finding it. The sonogram for #1 shows some curious faint mushy sounds at the same frequency as the double knock about a half second later; there seems to be a gap between the double knock and these mushier sounds. I wonder if these second sounds might be echos of the original sound, perhaps reverberating off the row of trees along the Obion. At this time the camera was pointed away from the source and towards these trees; echos from these trees would have arrived at the camera with about this much delay. Overall I suspect each of the double knocks was of similar intensity, and the differences in the recording are primarily a function of whether the camera was pointed towards them or away from them.

On March 17th, Scott, Melinda, and Ryan had all been together in the northwest part of the WMA while Scott was using the simulator. Melinda commented that in the entire morning's worth of simulations, only one single simulated double knock actually sounded like the real Campephilus sound, which she is familiar with from the tropics. That's one individual double knock, not one series. Both she and Scott have been struck by how much the ones I recorded, especially the loudest one at 7:01:09, do not sound like Scott's simulations and do sound like the real thing.

Is there a smoking gun somewhere on the tape that would confirm that they are not the simulations? There may or may not be. In listening obsessively through the rest of the tape, I've found what seem to be to be two very faint doubleknocky sounds earlier, at 7:00:11 and 7:00:31. They are exceedingly faint, and only Scott is pretty sure that he can hear them too. It takes headphones and careful listening to pick them out; the quality of the audio on the streaming video I have posted may not be good enough. I also believe that they do register on the sonogram:

This is an edited sonogram; in reality the sounds occurred 20 seconds apart, not 2 seconds apart.

One might wonder why I single out these particular nearly invisible dots for special treatment from among all the other nearly invisible dots. Well, it's because they are also associated with something I feel I can hear, so it's a judgement call. At or around 7:00:21 and 7:00:41 I don't hear anything, and there are no suspicious near-invisible dots on the sonogram. These two earlier faint "double knocks" have longer and more varied internal spacing than the three loud ones. Scott feels they are much more like the way his simulations actually sounded; they sound to me more like what I heard at 7:30. If Scott did his simulated series on time, then these two sounds are positioned just right to be the 1st and 3rd of his series. They would have happened at 7:00:03 and 7:00:23 by Scott's watch at his location, allowing for sound propagation time and the difference between his watch and my GPS. If these are real, they confirm that the three loud double knocks began right when the simulations should have been ending, and continued for an additional 20 seconds.

Finally from the 7:00 tape, there's this odd little sound:

It happens at 7:00:53, 4 seconds after the first loud (real?) double knock, and 42 seconds after the first faint (simulated?) "double knock." It comes just milliseconds after some splashing sounds almost surely made by my paddle. It has a quality much like a single woody knock. It would be just two seconds behind schedule to be the final simulation in Scott's series, assuming the other "faint" sounds are Scott, if he flubbed it and made a single loud knock instead of a double as happens very easily if you swing the striker too fast. Or... maybe it's just me klunking my boat with my paddle.

My recording from 7:30, with the camera pointed directly at the simulated double knocks, did not pick up anything definite. I do believe there is one of the simulated double knocks faintly resolved on it, probably the first one. Immediately after this sound, I swung the camera to the right. As I mentioned in the previous post, I was surprised that the simulations were coming from farther west (to the right) than I had expected. There's not much to see on the sonogram, but there are some smudges that seem to correspond to what I hear. The prominent vertical streak at the middle of the image is just a spurious "click;" the interesting sound is marked by the two arrows to the right of the "click:"

Here's a very small video clip of the 2 seconds of tape that contains this sound. You will hear the "click" prominently; the knock follows immediately and is less obvious.This also shows that the surface of the lake was still glassy calm, with not a hint of wind having developed yet:
To my ear this very faint sound consists of two rather widely spaced knocks, the second fainter and also higher in pitch than the first. This is something I noted about the simulations, that the pitch of the second knock frequently seemed higher than that of the first. This was not true of the loud knocks I recorded at 7:00. I have searched through the 7:30 recordings for traces of the remaining four knocks in this series. My ear picks up nothing at all. Sometimes the sonogram suggests the presence of something at about the right time, but it never matches anything audible and it is never all that distinct from many other small specks and smudges. Even the one I highlighted above is not certain in my mind; I picked it out particularly because of my own actions (pivoting the camera right after) suggest that I heard it and recognized it.

To get to the final point, in order for the double knocks I recorded to be Scott's simulations, ALL of the following things must have occurred:

1. Scott must have either miscounted his double knocks, doing more than just five, or have begun his series 20 seconds late.

2. Three of the double knocks would have to have propagated anomalously in such a way as to arrive at my location, 1800m away through the forest, sounding and looking as crisp and clean as if they were only a few hundred meters away.

3. The anomalous propagation would have to have somehow not affected the first two double knocks. There is nothing similar to the loud double knocks to be found at the times they would be expected, in spite of the camera having been pointed more nearly in their direction

4. The anomalous sound propagation had to be highly directional, so that Melinda heard nothing in spite of having a shorter path with more open space between her and the source.

Scott remains fairly certain that #1 did not happen, and the faint sounds early in the tape give some evidence that he did indeed begin on time. Even if he did begin late, the other three items require a coincidence of multiple unlikely events. It seems to me a more straightforward conclusion is that I recorded the MIMDKWFTII in the woods maybe 400m from me. The 10 second timing argues for the simulations, but everything else argues against it. When you take in to consideration that I heard an identical double knock series, including the short (10-20s) intervals between double knocks, originating from the same area just three weeks before, when there was no double knock simulator within 100km of us, then it doesn't even seem all that improbable. Of course, even if this is the case, this doesn't add any new evidence at all as to what the mystery double knocker might actually be. It just confirms that we Tennesseans have actually been hearing the "right" sound.

Other posts in this series:


At 3:38 AM, Blogger cyberthrush said...

some questions:

1) does anyone know from past study of campephilus species whether a 10-second gap between DKs would be at all typical (is that why Scott chose a 10-sec. gap?) , or is the spacing simply random in other species?

2) I'm guessing you're thought is that the DKs you heard may have been a direct response to Scott's simulation -- IF that's so, do we know if it would be typical of campephilus to wait for 5 repetitions before reacting in kind?
And finally, IF the bird(???) was responding that means it could hear Scott's DKs fine, yet Scott did not hear the return DKs -- does campephilus have far more acute hearing for DKs than humans do? or, given the positions would Scott's DKs have been much louder to the possible bird than the responding DKs would've been to Scott (my impression from all you say is that Scott's DKs are definitely NOT louder)?

At 5:58 AM, Blogger Bill Pulliam said...

All good questions, and will be subjects of much more discussion in upcoming posts. Later in the spring cott, Martjan lammertink, and I had a long e-mail dialogue about these very issues. In short:

The 10 sec spacing was standardized by Cornell. It was designed not because it is typical Campephilus but for convenience and recognizability. As for what real Campephili do, it seems to depend on species, season, etc. Almost no info on the details of this as pertains to Ivorybills.

About what sort of response, if any, to expect, again it depends on species, time of year, etc.

Final point echos the incident in 2008 when Scott thought he heard a response to a simulation he did not hear. It seems a woodpecker might be both more keyed in to the sound and better able to hear it from its elevated location as compared to our ground- or water-based spots. Just conjecture, however.

At 9:37 AM, Blogger Joel said...

Perhaps you have already made the necessary adjustments, but this sort of confusion could be avoided if Scott turned on his camera before doing his DK's and initially pointed it at his watch for a time stamp. That would give more precise timing but would also record the cadence of Scott's simulations and any potential responses.

At 10:49 AM, Blogger onthecoyle said...

In case anyone forgot or had not seen it, BBC's "Life of Birds" DVD has a nice clip of a double-knocking Magellanic Woodpecker. I don't how much production and editing went into the piece, but in that instance, the bird responded to single simulated DKs with single DKs (if I recall correctly).

Also, in that instance, the DK simulation was old David Attenborough rapping his fists on a tree trunk.

At 8:39 PM, Blogger Bill Pulliam said...

Joel -

I began doing this when I was operating the simulator. Before this incident I don't think we had really thought through fully what we would do about responses and how we would identify and document them.

Coyle -

I'll talk more about this later. Attenborough actually used two rocks to knock on the tree. Lammertink et al have a fair bit of experience with simulators and Pale-billed Woodpeckers.

At 9:31 PM, Blogger mabel said...

I recently came accross your blog and have been reading along. I thought I would leave my first comment. I dont know what to say except that I have enjoyed reading. Nice blog. I will keep visiting this blog very often.



Post a Comment

<< Home

Site Meter