Ah I'm obsessed with sound recording and spectrograms. It's like a whole new secret world have been revealed. Luckily we're on half term this week so I can devote the next 7 days exploring its wonders.
The only downside is that I'm neglecting the wife again. She was complaining last night about how I don't take her out any more, and I don't think I helped matters when I asked if I could record what she was saying to make a sonogram out of it ;-(
Right following on from my Sonogram/spectrogram success, and my new status as trainee boffin, a real boffin has been giving me advice on where to go next. I need to "...start comparing one call with another. How reproducible are the calls? Do individual birds vary their calls? Do the calls vary between individuals? Do these things differ for different species?"
Well here goes. Here's a Chaffinch which was right over my head yesterday:
And here's his song after it's been cleaned up, amplified and had other background birdsong removed:
And here's the sonogram for one of the bursts of song:
Now here is something very interesting. What does it sound like slowed down 5X and the pitch reduced by 5X? The wife was watching TV while I was playing with the clips (she will insist on doing that; Eastenders while I'm doing science!) and she heard this clip and thought it was a tropical forest bird from Venezuela. This is a digression from the science we are doing today, but what about the idea that small birds from Temperate zones have such fast metabolisms that they live 5 times more quickly than slow big Tropical birds, and in fact when one Chaffinch listens to another, what they really hear is this:
And if we speed up a tropical bird, does it sound like a Chaffinch? Another experiment to add to the list.
Ok, let's address the question "Do birds vary their calls?"
Here's the sonogram of 2 consecutive bursts taken from the above recording of the Chaffinch:
It looks like Arabic writing doesn't it? This is no coincidence; as I understand it, Arabic writing is a phonetic transcription of the pitch of the sounds (which is why their poetry looks so wonderful; another post!)
Listen again. Which sonogram is the sound represented by, the upper or the lower?
Ok, leaving aside the end part till later, let's look at the beginning and middle of the sonogram:
I've highlighted in purple, green and yellow the sounds which I think are copies, which leaves the "extra" ones in red. There seems to be an extra trill at the start of the burst, then the 5th sound doesn't seem to have an exact equivalent in the lower burst. And the final burst in the lower one seems to be different too.
First conclusion
Yes Chaffinches vary their song, even from burst to bursts, but where the bird does repeat its song, it seems to be an exact copy, at least down to the level shown in sonograms of this resolution.
How about Pitch and Speed? Does it vary? Lets blend the 2 sounds together (those nights of Photoshop training with with The Fat Capri are finally paying off)
The overall length of the song is remarkably consistent, in fact there is no apparent variation when comparing the start and finish time, taking into account the extra bursts.
In the centre, the 2 bursts starting at 10.85 seconds start at a slightly different time but the bird "catches up" by the time it gets to the next part of the song at 11.25, and there is no change in the actual speed of the song, only the spaces in between.
The maximium difference looks to me to be around 0.02 of a second. There is no difference in pitch and the only other difference can be seen on the burst which starts just after 10.6 seconds:
The falling note has a slightly different gradient but starts and finishes on the same frequency.
Second conclusion
The bursts of song are for the most part identical. There is a slight difference in the timing of the delivery but no difference in pitch or length of bursts. Slight differences in song are possible.
Now let's look at the end of the song:
Again they are remarkably similar, and any slight variation could be due to the amount of post-processing and magnification I've applied to the images. The first trill has 12 peaks in both samples, and lasts for exactly the same time. Each trill is around 0.005 of a second, and is still perfectly accurate!
Now I need to compare this to other Chaffinches
Comments