Moose Math
Added 2025-01-22 00:43:42 +0000 UTC
Comments
I feel like you should use one of the LUFS, I'm not sure how the LU range would be applicable in this case. Also, if you want to do statistics you will need significantly more repetitions with a consistent source like a loudspeaker. P.S. I really really like your videos and I hope you can continue doing what you're doing, it's amazing <3
Rubén Quiroz Marnef
2025-01-28 19:47:58 +0000 UTCI want to point out that LUFS as a unit were designed specifically for perceptions of loudness for humans. I don't know the anatomical similarities or differences for moose, so I have no idea how reasonable it is to use the unit for them, but I thought I'd mention it.
75 Million Inferior Goods
2025-01-25 00:25:03 +0000 UTCHaha no worries. I also just wanted to get ahead of the "well, actually..."s
Hendry
2025-01-24 16:04:50 +0000 UTCI hope you get some statistics advice. Unfortunately the best I can help? mumble out on this is um sounds plausible, eh, ugh, er resonance of the antlers to the frequencies of the cow call... bone conduction.. piezoelectric contact mic... I’d love to help more, but unless your problem involves composing pure functions with higher-order lambdas, optimizing tail-recursive calls, or debugging lazy evaluation in a curried monad stack, I might just end up being a side effect. Good luck with your quest!
Trey Dempsey
2025-01-24 02:32:07 +0000 UTCI guess the hypothesis is that the antlers allow the moose to both detect and identify the cow moose call. I'm starting to think there are good arguments for all three, I just wish I had picked one before I saw the graphs because I don't want to be biased
Hendry
2025-01-23 04:28:16 +0000 UTCI agree a series of tones would have been better in retrospect, I wanted to be sure to capture the "range" of the call, but that has brought with it new problems haha. Good idea with the audiogram for a visual, if just for fun. 1) Yeah I guess the best version of this experiment would have two mics and two moose ears maybe like 5 feet away from each other that both rotate together. 2) Yup lol. I don't have a stationary speaker that wouldn't need to be plugged in to an outlet and I drove to the middle of nowhere to find somewhere quiet enough in the first place haha 3) Yep! Makes sense I think I might show my results and then explain the ideal set-up for someone with more equipment in the future
Hendry
2025-01-23 04:15:03 +0000 UTCI think integrated LUFS (aka average loudness) is the most relevant measurement because it directly relates to your hypothesis. Your hypothesis is that the antlers amplify sound when ears are pointed towards them. Notably, (if I understand correctly) you're looking at the effect on sound in general, not specific peaks or short bursts. If the hypothesis is correct, one would expect a greater average loudness when the ears are pointed toward the antlers, which is what integrated LUFS measures.
David
2025-01-22 17:17:33 +0000 UTCThe choices to use two different microphones and to use a "complete" cow call seem problematic. 1. Having the extra microphone to measure ambient noise wouldn't help if it's not functioning exactly the same as the ear microphone, because they could e.g. capture different polar patterns. 2. You say you used a "cow caller". Is that one of those handheld whistle devices that require the user to blow through? If so, it would add a ton of variation, because humans aren't perfectly consistent. In that case I'd recommend prerecording a single whistle and playing it from a stationary speaker. 3. If the cow call uses multiple tones and pitches, Integrated LUFS seems the most sensible, because unlike Maximum Short-Term it would be increased even if just a few lower-volume tones get heightened, without the maximum changing. Points 2 and 3 would be easy to solve if instead of using a complex "cow call" you used a series of tones, from low to high frequency. But I admit those would not be moosey enough. For visualization of the results, obviously you should attach some of the recordings, but I think the most fitting graph would be an Audiogram! They display hearing level (in dB) by frequency (in Hz), they're used when testing hearing loss, e.g. to show the difference between normal hearing (horizontal line) and a partially deaf ear that hears "lower volume" (horizontal line that's lower) and a partially deaf ear that hears "low ranges only" (a "notch" in a particular band of frequencies).
Shemetz
2025-01-22 09:47:05 +0000 UTCYeah the mics were different unfortunately. I assumed the importance of the ear shape is in essence a tube that should increase the directional aspect of the mic, but maybe there is something more complex going on. I had two side variants with the antlers to try to capture the focal point. Both seem to show the effect. Good point about the MST LUFS
Hendry
2025-01-22 03:15:13 +0000 UTCHmmm interference is actually a pretty good point, like assuming the amplification effect is real, it could definitely affect the front as well
Hendry
2025-01-22 03:13:54 +0000 UTCFirst off I'm by no means an expert, just the ramblings of a backseat Internet scientist. With two different mics, especially when they’re of different types (one being more directional than the other), I imagine it’s always going to be challenging to confidently interpret the differences between them. Could the shape of the ear play any role in the results? I also noticed that the ear position in your setup looks different from the one shown in the paper. Do you think it’s possible that the antlers create a specific focal point? Based on your results so far, it doesn’t seem like there’s a distinct focal point, but it’s still something to consider. Regarding the measurement to use, I’d lean toward Maximum Short-Term LUFS. Integrated LUFS feels too generalized—it averages out the peaks and valleys, whereas the intensity of those fluctuations might be critical for detecting the call.
3volutionmedia
2025-01-22 03:12:04 +0000 UTCProbably Not what you are looking for, this make me things of sounds dispositive that vibrate the cranium to induce sound directly to the ear, given how antlers are attached, a moose brain could also "feel" the call directly. It's quite over my head but if negative value of LUFS means the sound loudness is reduced, could it be due to "interference" or even noise cancelation because echoing waves on the antler cancel the sound? Result could be extremely sensitive to position in that case. That s way more than rambling :o
Sven F.
2025-01-22 01:56:31 +0000 UTCI suppose repeating the experiment might be required. If I do I would absolutely get way more duplicates for variance. I was racing a thunderstorm and ended a bit early when rain started to threaten my electronics. Laptop fans were at least 3 feet from both mics
Hendry
2025-01-22 00:55:16 +0000 UTCI cant give any advice around #1. For #2. Obligatory: if variance is a concern repeat the experiments and find the relative error between them. If there's high relative error in variance between experiments then of course the experiment methodology should be reconsidered. Maybe the laptop fans between are getting louder or quieter between trials which messes with the noise floor? Worth getting more samples. I'd be interested if the trend still holds with more samples - it'd suggest the moose skull itself without antlers does some sound shenanigans.
Spamwhitch
2025-01-22 00:52:09 +0000 UTC
