song recognition apps

To: Allan Richardson <>
Subject: song recognition apps
From: Martin Butterfield <>
Date: Wed, 15 Oct 2014 16:44:16 +1100
I find this intriguing and suspect that Allan's answer is completely
accurate *if* you wish to identify every bird that one hears,

I have put that qualification because of my use of a Software app for iPad
(Shazam) that can listen to about 30 seconds of a piece of music and
identify it in many cases.  It isn't very good on classical music but on
pop and the sort of music on ABC Classic FM Drive  my guess is that it is
correct at least 90% of the time.  My guess is that there are several
orders of magnitude more CDs around than there are species of birds in
Australia (especially if seabirds are excluded).

Of course some birds have many different calls (without worrying about
mimicry).  But my guess is that matching a pretty high proportion (lets say
80% as a guess) of calls recorded in the bush to an authoritative source
(eg the BOCA tapes or a selection of Xeno-Canto calls) would basically be
quite achievable.

The feeling I get from coverage of this issue by a North American
researcher is that he wasn't game to proceed until he had got to a
situation of about 99% matching.   In other words the choice was perfection
or nothing.  I don't have the kit or knowledge to try this, but suspect
that if a greater level of "no match" was accepted something could be done
quite readily that would be way more useful in the field than the sample
calls in (for example the Morcom App.).


Martin Butterfield

On 15 October 2014 15:27, Allan Richardson <> wrote:

> Hi Michael,
> The short answer is no.
> What you’d like and what technology can provide in this case are miles
> apart.
> There is a recording device available that records bird calls and has
> software for reading the calls. However, you have to record, identify and
> then build a library of calls for the software to identify.
> However the technology is not reliable for a number of reasons, and the
> number of call variations you would require to cover Australia’s avi-fauna
> is beyond computation.
> The software has difficulty in identifying the calls, due to variation in
> intensity, pitch, shape and other noises masking the calls, so finding a
> good set of “standard calls” that the software could recognise would be
> difficult.
> Not only that, in any given area you may have as many as 200 or more birds
> that could be possible, and this multiplied by the number of call
> variations and loudness would be very difficult for software to work out.
> Hopefully a new way of deciphering calls will become available, but at the
> moment it doesn’t exist.
> The ability of the human ear and brain to identify the calls of birds is
> very difficult to emulate in the electronic world, due to the often subtle
> variation between species and within species vocalisations.
> I wish the answer could be of more help - the best thing to do is get out,
> hear a bird call and then track it down and identify it.
> Yes it’s slow and time consuming, and often frustrating, but these
> problems are what makes bird identification so rewarding in the long-run.
> Happy birding,
> Allan Richardson
> On 15 Oct 2014, at 10:28 am, Michael Hunter <>
> wrote:
> > Hi All.
> >
> >           Are there any Australian Bird apps which identify bird calls
> or songs in the field.  ie which “hear” the vocalisations and can identify
> the species which is calling?
> >
> >                           Cheers       Michael
> > <HR>
> > <BR> Birding-Aus mailing list
> > <BR> 
> > <BR> To change settings or unsubscribe visit:
> > <BR>
> > </HR>
> <HR>
> <BR> Birding-Aus mailing list
> <BR> 
> <BR> To change settings or unsubscribe visit:
> <BR>
> </HR>
<BR> Birding-Aus mailing list
<BR> To change settings or unsubscribe visit:
<Prev in Thread] Current Thread [Next in Thread>

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the birding-aus mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU