Sound recording and sonograms – an absolute beginner (1 Viewer)

I have always considered sound recording and sonograms to be the expensive preserve of a technically-minded few. However, as a result of reading other threads on birdforum (particularly this one - http://www.birdforum.net/showthread.php?t=65692) I have become interested in the use of these techniques for bird identification, and in particular for the identification of the different types / species / subspecies of crossbill.

I thought it might be of help to others starting out in this field to share my experiences as an absolute beginner, trying to find a way of sound ID without breaking the bank. Hopefully, more knowledgeable people may also be able to provide me with advice as I proceed.

I should state at the outset that I am not looking to produce high quality sound, just usable recordings. I am trying to do this on a budget, because a) I have limited time available to spend in the field so an expensive set-up could not be justified in terms of the likely outputs and b) if I (with my total lack of experience) can get a cheap set-up to work it might encourage others to take up the challenge and add to our knowledge of crossbills and other species.

The equipment

Obviously, first of all, I needed a microphone and some sort of recording device. After reading one of griffin’s posts I bought an Audio-Technica ATR 55 for about £60 including P&P. This is a shotgun microphone designed for use with camcorders and has two settings (telescopic and normal).

The recording device was more of a gamble. I have seen a number of suggestions on birdforum including old-fashioned tape recorders, minidisk players, ipods and flash recorders. Griffin’s advice was that it is important to record in the uncompressed wav format, as compressed formats such as mp3 may lose some of the frequencies required to produce a sonogram. Unfortunately most of the recording devices that I could find that recorded in wav were a bit too expensive (£200-300). I understand that some of the latest ipod models can record in wav but mine is a couple of years old so I could not use that. Eventually I found a cheap wav voice recorder (Denpa MP38) on ebay for £75 (new). Even though I couldn’t find any further information about it I thought it was worth a gamble at that price.

There is a photo of my set-up below. As you can see, it is fairly compact and portable (for opportunistic use if I am out for a walk). I reckon I will be able to record up to 4 hours of material at a time at the highest quality setting.

For sonograms I again followed griffin’s advice and downloaded the free gram23 software, which I found via google.

Yeah, that mic looks familiar, very much like my Monocor which is basically a rip-off of the ATR55.

Have you got any sound samples yet Paul ? Would be interested to hear how your recorder performs and how much gain it achieves. The great thing about Mini Disc recorders is that you can manually pump the gain right up with a mic like that, and although there will be some noise from the mic it will grab quite distant sounds like crossbill flight calls, clear enough for a sonogram. If it has an automatic gain setting you are at the mercy of the machine.

I need to do a survey of some "common" patches around Durris, Fetterso etc soon - you are welcome to tag along, and I can maybe give you some tips having also been at the "entry" level re-equipment myself ! Most of my prized recordings are so far with my "old" cheapo setup, though the Telinga is earning its keep nicely !

PS there are "parakeet" types at Durris just now and possibly Bohemian.

Linz

Right then - my first attempts at recording.

Given that I don't get crossbills in my garden, the nearest approximation I could find was a trilling greenfinch at the top of a cherry tree.

I recorded about a minute of trilling, imported it to my PC (the recorder came with a USB connector) and ran the file through gram23.

The result was something that looked like the interference between channels on a television. I messed about with some of the settings on gram23, settings like FFT, theshold etc. - no idea what they mean. Some smudgy fingerprints appeared on the sonogram (see first picture below), but still not very good.

At this point, a little disheartened, I decided that I should try to edit my sound file and pick out the clearest section. I don't have any dedicated sound editing software, but I discovered that I could load the wav file as an audio track into my camcorder editing software (Ulead video studio), cut out the best bit and save it to a new wav file. At the same time I upped the sound levels a bit in the video editor.

I'm not really sure what the video editor did to the wav file, but when it came out the other side, my 700KB one minute recording had become a 2.2MB 13 second recording. Apparently the bit rate (whatever that is) of the file had increased from 88kbs to 1400kbs.

Anyway, without really knowing what I had done, I ran the new wav file through gram23 and amazingly this time I got a recognizable sonogram (see second picture below).

The bursts of 3 - 6 vertical lines near the bottom of the sonogram represent the trilling of the greenfinch. The sloping lines and other marks above these are some coal tits that flew past as I was recording.

I don't know if this sonogram is any good, but I was pleased to at least get something at this early stage. The bird was about 10-15 metres away as I was recording and I certainly managed to get this close to calling crossbills at Fetteresso and in the Forest of Birse during the summer.

Linz, there is a mic gain setting that you can adjust on the recorder - I can't remember how it was set when I made this recording.

I would be interested in coming on one of your survey visits if I am available - nothing like learning from an expert.

Capercaillie71 said:

At this point, a little disheartened, I decided that I should try to edit my sound file and pick out the clearest section. I don't have any dedicated sound editing software, but I discovered that I could load the wav file as an audio track into my camcorder editing software (Ulead video studio), cut out the best bit and save it to a new wav file. At the same time I upped the sound levels a bit in the video editor.

Click to expand...

Nothing wrong with using a piece of software that does the job, but if you were after a dedicated sound editor the open source (free) editor audacity is pretty good

http://audacity.sourceforge.net/

I'm not really sure what the video editor did to the wav file, but when it came out the other side, my 700KB one minute recording had become a 2.2MB 13 second recording. Apparently the bit rate (whatever that is) of the file had increased from 88kbs to 1400kbs.

Click to expand...

You don't say what model recorder you have but I guess this is a voice recorder, which btw is dead useful for field notes - I use one myself. These usually sample at 8kHz, which will restrict your audio frequency range to 4kHz, and usually compress the audio to get it on the chip. That's not totally optimised for bird recording.

The shortening of the duration might be ulead interpreting the file as a 48kHz sampled file, which is what it needs to author a DVD. Does the sound still sound like a greenfinch when you play the 13s clip?

ermine said:

Nothing wrong with using a piece of software that does the job, but if you were after a dedicated sound editor the open source (free) editor audacity is pretty good

http://audacity.sourceforge.net/

Click to expand...

Thanks for that link - I have downloaded and it looks just the job.

ermine said:

These usually sample at 8kHz, which will restrict your audio frequency range to 4kHz, and usually compress the audio to get it on the chip.

Click to expand...

I have no idea about the frequency range - the instructions have been translated from chinese with limited success. However, it does claim the ability to record as uncompressed wav files as one of its main selling points:

http://cgi.ebay.co.uk/DENPA-MP38-DI...ryZ62041QQssPageNameZWD1VQQrdZ1QQcmdZViewItem

ermine said:

The shortening of the duration might be ulead interpreting the file as a 48kHz sampled file, which is what it needs to author a DVD. Does the sound still sound like a greenfinch when you play the 13s clip?

Click to expand...

Sorry I don't think I explained it very well in my last post. The shortening of duration was because I had saved a 13 second extract from the original recording. What I didn't understand was why this extract was 3 times the size of the original one minute file. It sounded perfectly alright.

I have attached another attempt at a sonogram from a 10 second extract.

Hi

Can't offer any advice but look forward to hearing what you get up to.

best wishes
James

Capercaillie71 said:

I have no idea about the frequency range - the instructions have been translated from chinese with limited success. However, it does claim the ability to record as uncompressed wav files as one of its main selling points:

http://cgi.ebay.co.uk/DENPA-MP38-DI...ryZ62041QQssPageNameZWD1VQQrdZ1QQcmdZViewItem


Sorry I don't think I explained it very well in my last post. The shortening of duration was because I had saved a 13 second extract from the original recording. What I didn't understand was why this extract was 3 times the size of the original one minute file. It sounded perfectly alright.

I have attached another attempt at a sonogram from a 10 second extract.

Click to expand...

This link

http://www.spyemporium.com/digital_audio_recorders_voice_denpa_mp3_player_long_recording_mp38.html

indicates that this records to wav ADPCM compressed, which would explain the 3x expansion I guess. ADPCM compresses by saving the difference between one sample and the next. Audacity will tell you the sampling rate and compression type when it opens a file - the highest frequency you can record is half the sampling rate.

The increase in bit rate when saved by ulead, probably uncompressed supports the ADPCM compression. The bit rate of 1400kbps is typical of CD rate WAVs (sampling rate x bits per sample x number of channels / 1000 = 44100x16x2 = 1411kbps)

If you choose your section using Audacity then you can adjust the volume of the clip so the peaks go as high as they can without maxing out (ISTR Audacity has a normalise function which will do this automatically). That will set the loudest bit, which corresponds to the black bits. You then tinker with the threshold control to stop the the background noise making snow on your picture - with a clear recording you ought ot be able to set this to about 40dB. It is a trade off with cutting off fine details and losing everything in the mush - you have to fiddle with it to see what works for your gear. You probably want os leave FFT at the default until you get more used to what the software does. Adjusting your clip for full level in Audacity and adjusting threshold in gram should give you clear sonograms if you have a clear recording. You're on the right general track - experience and fiddling with the threshold variable will get you good results.

ermine said:

You're on the right general track - experience and fiddling with the threshold variable will get you good results.

Click to expand...

Thanks for the help ermine. Reassuring to know I'm on the right track. It will be interesting to see how well this set-up works at greater distances from the birds.

Yesterday I received my copy of 'The Sound Approach to Birding' by Mark Constantine

http://www.soundapproach.co.uk/

I would reiterate what Linz said on the crossbill thread. This is a great book and even for a beginner like me, it is easy to follow but packed with information. Anyone interested in this subject should buy it if they can.

I used the "telescopic" setting on the mic at all times. As for the mic gain try the recorder on the highest ( most sensitive) setting and see how you get on. It may introduce more noise but will help pick up faint chippy crossbill calls. This setting will aslo introduce more handling noise as well though.

Linz

Capercaillie71 said:

It will be interesting to see how well this set-up works at greater distances from the birds.
.

Click to expand...

You're probably pushing the envelope of a voice recorder with agc at much greater distances than 10-20m. You just have to give it a go and see if it pans out, but if you find you can't get enough level out of the recording in distant situations where you can still hear the birds then you might want to see if you can borrow a MD from somewhere...

Yesterday I received my copy of 'The Sound Approach to Birding' by Mark Constantine

http://www.soundapproach.co.uk/

Click to expand...

I had a chat with Mark at Birdfair, they get up to interesting antics on the Dutch coast. I hadn't realised you could pick up flight calls with a parabolic dish even if you couldn't hear them unaided, I will give this a go on the east coast this autumn.

Since you have the book, you can record some of the CD tracks and play with the sonogram software there. You then can see what the sonogram should look like in the book and find what adjustments of thrshold give you similar results.

ermine said:

I hadn't realised you could pick up flight calls with a parabolic dish even if you couldn't hear them unaided, I will give this a go on the east coast this autumn.

Click to expand...

Crossbill flight calls at 200 yards = black on sonogram with a Telinga Pro6 Twin Science. Got a Parrot at Abernethy with my ME67 and it was same distance but not as strong. Recorder both times Fostex FR2, incase anyone is interested.

With my cheapo Monocor I can ID flight calls on sonogram up to 100-150 yards depending on conditions when Mini Disc recorder is used. Some experience needed in reading the sonogram though as vital components are often missing or faint.

Great for "flyovers" which for most people is what crossbills are. At a native pinewood on Sunday ( with Telinga ) I didn't even see a crossbill but recorded Scottish and Parrot flight calls, and some Parrot song. If the birds are moving about in the canopy it is often difficult to see them when you are under a blanket of it. That said it was unusual in that I often get good looks at them, though this particular site is reknowned as being "stubborn" and is easy to blank it.

Linz

Capercaillie71 said:

I would be interested in coming on one of your survey visits if I am available - nothing like learning from an expert.

Click to expand...

Not sure I would call myself an expert ! I am certainly not a technical geek that is for sure. However, when it comes to sniffing out crossbills
(anywhere) I don't fail (often), so perhaps a 'crossbill finding expert' ? Will let you know the next Sunday that is available to come along - I ring on Sundays though so it will be weather dependent.

I am calling Magnus Robb ( of Sound Approach ) sometime in the next week when he gets back from recording Stormies to ask his advice on something and he also considers himself "untechnical" and instead relies on his ears ( as a musician like me ) and his intuition and fieldcraft in getting the birds. Basically he goes out and "does" it. He recommended the Fostex FR2 ( which he also uses ) even though I had heard negative views elsewhere about it's suitability for bird recording. A listen to one of his recordings will completely refute such claims. Formula's and bumf and mic specs etc don't always equate to producing a good recording ! Also, all equipment, including Telinga mics, have limitations and it is being able to work within these limits that is the real skill.

I am really pleased you are having a go and hope it will encourage others to do so, especially all the whingers that say Scotbill shouldn't exist because they can't ID it !

Keep working on the Greenfinches as some of their chipping is very similar to crossbills ( so much so that people mistake them ). Chaffies are also good to practice on and although the cadence of their flight calls is slower than a xbill, it is about the same intensity so will give you an idea of distance effectiveness of your set-up.

Linz

I've been playing around in my garden again. It was a bit windy and I had to make do with the small number of birds available, but here are two more sonogram attempts.

The first sonogram shows two bursts of blue tit at about 20 metres (the first two bursts of vertical lines) and a final more smudgy mark of a jay that called about 50 metres away (quite rare for my garden too).

The second sonogram shows jackdaws calling about 250 metres away.

In both cases I had the mic gain at maximum and amplified the file in audacity, as ermine suggested.

You're struggling a bit for signal to noise ratio. You need to move closer to the birds or choose louder birds. The wind won't help - there are things you can do to help with that. A sock over an old mesh peanut feeder, with the mic held in the middle with rubber band so it doesn't touch the sides can lose most of that.

A blue tit at 20m away is not a loud noise. It is a lot less loud than close speech which is what a voice recorder is engineered to handle. I do feel that if you are after working at that sort of range with such low level signals you need a better recorder. From the looks of your sonograms you have learned how to control signal levels and adjust threshold adequately.

You'll find it easier when things aren't particularly windy. What you need is a bunch of friendly and vocal Sparrows, then make some recordings at different distances from them to get a feel for range.

Getting clear recordings at a range further than you can easily see in 8x binoculars is always going to push the capabilities of recording equipment. Do you have a camcorder with a mike input? Try using that to record - the audio signal quality of MiniDV is rather good. Don't use the inbuilt camcorder mike - the noise of the motors which wipe out any bird sounds. You can also try putting the mic into your PC on a long lead and using Audacity to record directly to the PC from your backyard birds with the mic pointed out of the window, see if this gives you a better range?

For a comparison, this is a sonogram of a Curlew making that lovely bubbling call, and the settings I used. The bird is about the same distance as your Jackdaws. Wind was about 25mph and the mic was a Sennheiser MKE 300D which is generally similar to the performance of yours

That's a nice clear sonogram ermine. If and when I catch up with some crossbills my main aim will be to see if I can at least get something that is identifiable. You may well be right that I will struggle with the voice recorder, but it will be an interesting learning curve.

I haven't been out this weekend as I have had a bad cold, but hopefully I will have a chance before too long.

Bummer about the cold. The curlew wasn't a great recording BTW - that is the one starting 46 seconds into this

http://www.suffolkbirds.co.uk/audio/060902_melton_curlew_600HPFord3.mp3

and the signal to noise is dreadful - the wind was 25mph and there's a busy road about 100 yards behind.

I really struggled looking for crossbills in Scotland April this year. I failed totally to connect with one in the Abernethy forest where they are meant to be two a penny. Finally heard one faintly in high winds by using a parabolic dish in the Culbin forest which was so far away I had to use a scope to see it at the top of a tree.

Just got my first useable recording. I use Cool Edit Pro 2 which produces a decent colour sonogram, attached. Bandpass filtered at 2kHz to 16kHz, and image sharpended slightly with freq up on RHS.

My first bird, however, is a bit of a mystery. I suspect it one call of the many faceted Great Tit. I uploaded the MP3 to HERE and would appreciate an ID if poss.

Kit : 32" parabolic dish + stereo dynamic mics into iRiver H120 recording in WAV format.

Crits welcomed.

Ta,

g8ina said:

Just got my first useable recording. I use Cool Edit Pro 2 which produces a decent colour sonogram, attached. Bandpass filtered at 2kHz to 16kHz, and image sharpended slightly with freq up on RHS.

My first bird, however, is a bit of a mystery. I suspect it one call of the many faceted Great Tit. I uploaded the MP3 to HERE and would appreciate an ID if poss.

Kit : 32" parabolic dish + stereo dynamic mics into iRiver H120 recording in WAV format.

Crits welcomed.

Ta,

Click to expand...

I go for Great tit. Has the curious "hee-haw" pace and rhythmn characteristic of this species.

I've attached a sonogram of one I recorded earlier in the year which shows distinct similarities. Indeed, to my eye they look more similar than they sound similar...