How to clear up my spectogram?

[Steelflight]

I have a spectogram that looks kind of like a long blob... there is supposed to be some crow cawing and seagull sounds in here. there is a whole bunch of st uff between -0.5 and 0.5 on the y-axis... and then some various spikes. how do i know what to edit out and how to i cut out this middle 'noise'?

im using audacity btw

[Steelflight]

ok i did 'select all' and then i did both a 'low pass' and 'high pass' filter. i left the settings for default-- is that ok? not really sure what freq to filter out...

still trying to clean it up a bit. keep in mind it was a bit windy and i did have some ambient static (not sure... maybe I should buy a Rycote wind tag--> the wind was maybe 5-10 mph tops).

for some reason when i try to post an attachemnt it just says 'loading' and takes like 20 minutes (maybe because it's a pretty big spectogram). anyways, any advice on editing out ambient garbage? what types of things do you guys/gals run in Audacity other than the filter options?

thanks!

[ermine]

You probably want to choose easier subjects to start out with. With an ME66 you will need some sort of wind protection in a 5-6mph breeze.

You can try a high-pass filter at 300Hz to filter wind and traffic noise, but if the wind noise has pushed the mic or recording over the top then you will not be able to filter it out post-recording.

Starting out, you will have an easier time if you record close to a vocal bird like a finch or a chickadee. I've heard Amercian birds are more approachable that European ones but I still reckon you won't get that close to a crow!
Audacity has the advantage of being cheap and is fine for editing and trimming. Its filtering capabilities aren't that comprehensive or easy to use. That's not such a bad thing - capturing a better signal in the first place by getting the recording levels right and getting close to the subject trumps any amount of filtering heroics (http://www.wildlife-sound.org/equipment/newcomersguide/eq.html). It's hard to say what's up without hearing the recording, if you can upload the section to a website I'll take a listen.

[Steelflight]

do you think a rycote wind tag would help? any suggestions on how to improve the wind-proofing other than the mzw?

thanks.

[Steelflight]

so far the only places i can find with a rycote wind gag are on the east coast... and im hoping i dont have to wait a week! could i go to like a fabric store and buy some fur and slap it over the mzw foam windscreen?

[ermine]

You can DIY wind protection, eg

http://www.joelandkaren.com/mic-zeppelin/

A rycote windgag on my Sennheiser MKE300 is good for about 8mph winds which is still a bit limiting. You'd be better off with a Rycote Softie or something like a Reinhardt Whisper - Linz reported good results with that and a ME67 http://www.reinhardt-film.com/whisper.php

The other option is to concentrate your recording forays on less windy days, at least while you're learning the ropes. Even with full wind protection on the mic. windy days aren't that great for recording as you will pick up the rustle of reeds, leaves and other stuff. Recording is a skill which you get better at with experience. You can often use the terrain to shelter yourself and your mic and gain advantage that way.

[lachlustre]

In answer to the question:
what types of things do you guys/gals run in Audacity other than the filter options?

Have you tried playing with the spectrogram settings themselves? In audacity, you are more or less limited here to the FFT size, and the reason I ask is because this command is rather hidden away in the preferences (at least in the Mac OS version). For the sounds you describe, FFT size will make a dramatic difference. The reason is that crow and gull calls have strong harmonics, which will not be very easy to distinguish if you have too small an FFT size. The longer the FFT size, the better the frequency resolution, and the worse the time resolution, but given that the calls you have recorded are relatively slow and unmodulated (compared to eg a warbler), you can get away with much longer FFT sizes.
Sorry if you know all this already!
By the way, to make really good looking spectrograms, I prefer the free program Praat (google for it).
I also have my own sound analysis program which incorporates a simple reverberation attenuation algorith. If you post the sound recording here, I'll post the spectrogram I can make with it.
Finally, I've never been impressed with Audacity's noise removal algorithms. But I've been reading some interesting bird song research recently which made use of the noise removal algorithms in the shareware program goldwave and the free (I think) program Syrinx. They might be fun to try out!
Cheers
Rob