Peter,
You are quite right to be suspicious of the effects filtering may
have on a sonogram. So let's go back to the original.
If you look at the left hand channel of the original recording in
Raven Lite, with the settings at 50 darkness and 50 contrast, and FFT
size 2516, that may help. Expand the sonogram window so that you are
seeing about 1 minute of sound, and only up to 10 kHz, then you
should be able to see two more harmonic bands at around 3 khz and 2
khz for at least some of the calls. At 1 kHz things get messy
because of other things calling.
In Sonic Visualizer, I could not find where to adjust the brightness
and contrast and FFT for the sonograms, and therefore could not get a
very clear result.
Izotope RX gave a good result (but only very slightly better than
Raven Lite), i.e. just looking at the spectrogram window of the
original recording and adjusting the controls for clearest settings.
You are right that normally it is the higher frequencies that are
attenuated by distance. In the case of the frog distress call
though, the strongest part of the call is not in the lowest
frequencies, but higher up, say above 3 kHz. So with Tom's call
being so faint in the first place, maybe the lower frequencies simply
did not pick up on the recording.
Also, I don't know what other effects the mp3 format may have had on
the recording.
cheers,
Vicki
On 04/12/2011, at 12:56 PM, Peter Shute wrote:
> Interesting. I can see the same sonogram patterns using Raven and
> Sonic Visualiser using the track you just uploaded, but on the
> original and all other attempts at cleaning it up, the sub 4kHz
> bands aren't really visible. Even in yours, they're much fainter
> that those above 4.
>
> I agree it looks like a good match, and a very likely one too,
> given the location, but I'm wary of something that's completely
> invisible on the original.
>
> Sonograms are a new thing to me, so I'm right out of my depth here.
> Do you think the distance and reverb can explain why the lower
> harmonic bands are fainter? I would have thought higher frequencies
> would be attentuated by distance more than lower ones (but I'm not
> sure about that).
>
> Or perhaps they're just almost completely masked by the frog
> chorus, and would have to be fainter once that's removed. On closer
> inspection, I can see a faint band around 3kHz on the original in
> a couple of spots (eg 28s), but I just couldn't say below that.
>
> Peter Shute
>
> ________________________________
> From:
> On Behalf Of vickipowys
>
> Sent: Sunday, 4 December 2011 11:25 AM
> To:
> Subject: Re: [Nature Recordists] Re: Advice needed for cleaning up
> this recording
>
>
>
> Peter,
>
> I'm sorry you've given up on the mystery call. Here is one last
> attempt on my part to convince you the mystery call really is the
> distress call of a Green Tree Frog.
>
> I've selected just a short side-by-side comparison, using the
> clearest part of Tom's recording that I could find. I've also
> presented the recordings at half speed, which is always useful for a
> listening test.
>
> This is the soundcloud link:
>
> http://snd.sc/ticMjy
>
> I've included a Raven sonogram that shows how the harmonics, although
> faint, do extend well below 4 kHz (you thought they did not).
>
> I did some broad band noise reduction on Tom's original recording
> using RX, and removed the prominent insect call.
>
> cheers,
>
> Vicki
>
> On 03/12/2011, at 7:26 PM, Peter Shute wrote:
>
>> Thanks everyone for your attempts at cleaning up this recording.
>> We've given up on identifying the call for now. I assume it must be
>> a lesser known call that we have no samples of for comparison. We
>> had quite a few suggestions that sounded similar, but nothing with
>> a matching sonogram.
>>
>> Now I just have to try to understand the steps you all took so I
>> can try for myself next time.
>>
>> Peter Shute
>>
|