A bat call library with Martyn Cooke

S3E29 This week Steve joins Martyn Cooke outside a holiday cottage in Staffordshire as they set up a matrix of bat detectors to record the sound of Brandt's bats emerging from their roost before flying into the adjacent woodland. Martyn explains why he is recording the calls, why he's travelled all the way from Surrey to record these bats and what the calls will be used for. Steve finds out how an injured serotine bat got Martyn hooked on bats and they talk about the advances in automatic bat call identification.

If you're interested in finding out more about bat detectors, your local bat group are likely to be able to lend you a detector and you can learn more about them on our website here.   Listen to our earlier episode from series one about Brandt's in the Churnet valley here.

• Follow Martyn on twitter
• Borrow detectors from us to help us grow the call library
• Bat Detector Info
• Low cost AudioMoth detectors on twitter

Please leave us a review if you can, it helps us to reach a wider audience so that we can spread the word about how great bats are.

Join the conversation on social media using #BatChat

Producer: Steve Roe @SteveRoeBatMan

Transcript

Steve Roe: 0:09

Hello, and welcome to BatChat. This is the podcast where we bring you the stories from the world of bat conservation. Today we continue in this current series with the bat worker who is creating a bat call library, which will become available to the public. I'm Steve Roe, a BCT trustee. And if you're a regular listener, it's great to have you back with us. And if this is your first time listening to BatChat, welcome along. Episodes are being released every second Wednesday from now through to the spring, and you can join the conversation online using the hashtag BatChat; that's all one word. As we meet each of our guests, you'll hear from people working to make a difference in the world of bat conservation, people who care about individual species, people who concentrate on one particular part of bat ecology, and people who are working with bats at landscape scale. As well as keeping up with the latest news and hearing from people in the world of bats. We hope that you'll be inspired to get involved because bats need our help. In this episode, we're with someone who specialises on one particular part of bats ecology, specifically echolocation calls. Today, I'm back in the Churnet Valley. Our guest today Martin Cook was visiting the area all the way from Surrey to make some recordings of one particular bat species. So I joined him as he set up his recording equipment, and got to learn more about what he's up to. My first question to Martin was what sparked his interest in baths?

Martyn Cooke: 1:31

Well, I was always a keen birder ever since I was sort of like seven or eight years old. But I mean, I was aware of bats, but I didn't really know much about them. And then rounded by 1997 1998, something like that. I found a grounded serotine. In the garden. Yeah. And in those days, well, my first reaction was this bat is absolutely huge. And it was a lot bigger than the pipistrelles. I was used to say, yeah, so in those days, there's no internet and no helpline and things like that. So I ended up going to the library to get a book out on the on British bats to identify it suddenly died. Well, that kind of sparked my curiosity. And then shortly after that, I saw a lecture advertised by David King who used to run backbox. And so I went along and I met a few people there that were in Surrey back group. And the rest is history is this.

Steve Roe: 2:43

And was it that meeting with David King that got you into what you now specialise in which I guess it sounds

Unknown: 2:49

Not really no, that was that was a guy used to work for BCT guy called Colin Catto. And I always remember Colin contacting me and saying, I've just got this fantastic new little detector. And it was one of David's it was a duet when it first came out. And he said it means you know, we'll be able to afford to, you know, record as it wasn't as frequency division, but we can analyse the calls. So he organised sort of an evening at the Surrey Wildlife Trust, and training centre at Norwood. We went along and we were recording bats over the ponds and in the woods there and back into the centre, and then looking at another computer, and it was just a complete, absolute revelation to me that you could actually do that. And I got hooked. And it's, you know, I've been doing it ever since. So why are you recording tonight's bats; we've put out something like 14 bat detectors on tripods scattered around this cottage that were stood in front of here? Why so many detectors and what are they for? Right? Okay, well, what we're trying to do tonight is record what we call voucher calls. And that is when you go you have a known bat. And then you record it so that you're certain that that recording came from a particular species about so here they are cottage here, which is a known Brandt's. Yeah. So by recording the bats that are actually leaving the cottage here, we know that, you know, the those are Brandt's recordings. And then we can use those for from helping us to do manual identification of bats, but also for training data for automatic classifiers.

Steve Roe: 4:46

And what is the end aim of the project?

Unknown: 4:48

Well the project I'm doing at the moment. I mean, it sort of developed from a project that I've been doing since 2017 Back in 2017 myself and Kate Jones from UCL went to the States. And we attended a week long, but echolocation symposium in the States. And one thing that was a recurring thing at that meeting was that large parts of the world, including the UK didn't have proper code libraries, where people can actually go to the library, download calls of known bats, and then use them. So that's what we started to do. So I started a project back in 2017. And travelling around the UK, recording all 17 species about we've got that has kind of been know passed on to BCT. And BCT, are actually going to produce the website that will host the call library. And it's going to be called EcoHub. And hopefully, well, Ray, you know, conferences in October, hopefully, by October, you know, we'll have something concrete that people can actually look at. That's gonna be my next question when it went out. So I mean, why do you record bats coming out of the roost are the calls not more atypical compared to what the users will be recording? Yeah if you, if you were to stand directly underneath the exit point, and record the bats there, what you'll get as the as the bats coming out of the roost, they require a lot of information quickly. Yeah, but what is going on in the immediate vicinity of that raised. So they tend to use very short duration, quite large bandwidth calls, which are not necessarily the same type of calls, they use what we call search phase calls when they're out foraging. And it's really the search phase calls that we want. So as I said, if you if you were to stand directly under the roost, the calls that you recorded would be sort of non typical calls. So what we do is we, we start off about 10 metres away from the roast and kind of work backwards. And what we try and do is find out where the bats actually root when they leave the roost. And then set detectives up along the along the way. And if we can, we try and pick spots, say a spot in the open a spot where the vegetation narrows down, so we get different different types of calls as well.

Steve Roe: 7:23

So that says the bats change in the environment.

Martyn Cooke: 7:25

If you're really lucky, you can you can get some quite nice transition calls, as the bats transition from quite a long duration, sort of open habitat call into a short, shorter duration sort of clutter type call. And that those are ideal, really, because it gives us a lot. I mean, the object of the exercise, at the end of the day, is to try and record the entire repertoire, if you like, of what that body's capable of doing. Not saying we're there yet. But you know, we're a good way into listening.

Unknown: 7:57

And we should say we're recording Brandt's, perhaps why have you come all the way up to the Midlands to call these do you not get Brandt's down south? Well, we do but that. I mean, I'm quite happy to say it. I think Brandt's bats are the rarest Myotis. But in the SouthEast of England, by a long way. I mean, we have far more Bechstein's we have far more Alcathoe, than we do of Brandt's and I mean, I'm a member of Surrey bat group. And we've been doing small Myotis bat surveys now for probably about six years now. And we're lucky if we catch, say three or four Brandt's a year. They are so rare, and particularly in Surrey, and I think Sussex is the same. We don't know of a single maternity roost in Surrey, we've had one roost that was found by a consultancy firm. And they sent droppings away that they found in this building. And they were DNA tested. And they came back as Brandt's. But there were no bats there. And there were no bats there again the following year. So we still haven't found an active Brandt's roost at all in the entire county. Is the software going to be able to tell the difference between whiskered and Brandt's bats? And how does it make the differentiation between such minute changes between those two species? Well, I mean, there's there's two things I mean, as well as doing the work for BCT. And UCL, I've also been helped develop the Sonobat software. And for that we measured 104 different parameters. So if you think you know, going back a few years when we were using some of the really kind of basic sound analysis software, we could probably measure the start frequency the entry can say peak frequent frequency call duration interval between the calls, you know, and then you starting to really push it so five, six parameters that you could measure and we know measuring Have over 100 different parameters both in sort of frequency and time domains. And UCL are also developing some software at the moment that I've been helping with, it's quite intriguing, they've got a completely different route. I think that they are the first people have actually tried this, but they're actually using pattern recognition software. So effectively, like face recognition type software, to to do their analysis with. But that has got a lot of sort of hurdles, they've got to jump to get that to work properly. And is that because the technology is newer, or it's just so complicated? It's just so complicated, because the big problem is when you record about is that your recording isn't necessarily what the bat was actually saying in the first place. As the the sound from the bat, the bat, makes the call, it travels through the air to your detector, your microphone, and the higher frequencies start to attenuate in the air. So what we ended up with a sort of fragments of calls where the top is missing, or quieter part of the bottom of the call is missing. And it's trying to get all these little subtle changes, not only in the bats, but also in the recordings as well. So it's going to be extremely challenging. I think, you know, to record a to classify bats to genius level. So Myotis bats and pipistrelles tell us should be fairly straightforward. But the pattern recognition software to start differentiating between, say whiskered and Brandt's is a big challenge for them. And you mentioned, microphones and detectors there, what are these detectors? You seem to have two different types of files these different tripods? Yeah, well, the ones are mainly generally because they're, they were good quality and relatively cheap. I use Peersonic rpa 3 detectors, which means that I could buy sort of six of those for the price of one more expensive detector. So it means that I've got six more chances if you like to record the bats. And it also enables us to same put the detectors in these different areas. We're also trialling the audio mouth, which is the sort of $49 detector that was developed at Southampton University. And so we trialling those to see how they compare with the other detectors. And I'll show tonight I've got a couple of USB microphones connected on to tablets, made by Pettersson. And so and there's there's quite a new one that I didn't actually use it tonight. But there's a brand new one just come out by Deltatronic, the Italian manufacturer. And that's the first one that will come in at under 200 pounds, which is quite a breakthrough for the USB microphones because they tend to be very expensive. And you've got 14-odd detectors here in on a typical night, I suppose it turns on how many batteries put on a typical night, how many calls so you genuinely recording? It's difficult to say. Obviously, a big thing is whether I've managed to put them in in the right place. Which you know, it's not a given. But I mean tonight, I mean, I don't know there was well over 100[bats emerge] a few nights ago. So I'd be disappointed if we didn't get sort of at least sort of eighteen recordings per detector. Yeah. But we'll just have to see. And how long does it take you take for us to go through those recordings? Are you going through and verifying those recordings, or is there software sort of doing stuff, Particularly with the audio moths. Because the audio moths don't have a what we normally describe as a trigger, most sort of commercial bat detectors have what they call an amplitude trigger. So as soon as the sound comes in above a certain amplitude, it'll switch the detector around, it'll record for a set period of time and then switch off again. Unfortunately, the audio moths because of the processor don't have that facility on them. So the way we've got them set up for the BCT project is that we just record for five seconds, and then it downloads that five second recording to the SD card that's built into them for one second, and then we repeat the process. So we're recording every six seconds. But what he does mean is that I've got six audiomoths tonight and they'll be 1200 files per detector. And the vast majority of those won't have any bats on them at all! So just take a while to go through them. And for listeners at home are thinking, oh, I want to get better recordings, or they get their detector recordings home. And there's lots of background noise, you know, what makes the perfect recording, and what should people do to get the cleanest recordings? The biggest thing is reflections. So, ideally, you want to get your detector or your microphone away from any sort of hard surfaces. And ideally, that the probably the best way of doing it, is to put them on a pole, or on a tripod, around about two metres off the ground. And that means that any reflections that come back will actually be displaced in time from the actual call coming direct from the bat. And that will dramatically improve your recordings. When you may, one of the things I brought with me tonight, I've actually got a backpack mounted detector with a pole that sticks up out of the back of the backpack. I mean, you look a bit silly walking around with it. But there is logic to it. Because your body acts just the same way as a treatment ward or a car would you get lots of reflections coming off your your off your body. And if your microphone is close to that, the the reflections meet more or less exactly the same time as the call from the back. And you get this sort of mushy trail of echoes behind the calls. And really, particularly for the classification software, we really need the cleanest best calls, we can get to give the classification software a fighting chance of being able to come up with a correct ID Which is why you've got all these detectors on tripods. Yeah, well, plus the fact it actually makes them easier because I say I can just place them where I want to place them, rather than having to try and find branches on trees that are in a suitable position. So first, when I was setting them up, I've got some small tripods and some larger, which are actually shoot photographic studio lighting stands that you can get. So it's just a tip to get the detectors and the microphones away away from the ground, basically. So you're hoping that this will be launched in October. I guess the million dollar question is, how far into the future do you think it's going to be before manually identifying bats through human is a thing of the past how long before we're all doing auto analysis? I don't think I don't think manual. Certainly manual vetting, I prefer to call it will ever kind of go out of fashion. Because with the best will in the world the computer algorithms will throw up spurious recordings. I mean, one of the classic ones are things like social calls, feeding buzzes, certain types of cricket, if you're not careful that the software thinks that that's about and it will try and he'll try and classify it. So there's always going to be a need for people to manually go through their record, thought it's not necessarily all the recordings, but certainly some of them to make sure that the classifiers are actually classifying proper calls for want of a better word.

Steve Roe: 18:34

And what do you like about and you've been doing for years? What do you like the most about the project?

Unknown: 18:39

Well, I mean, I think, you know, it's something that hasn't been done before, certainly in the UK. So that is quite challenging. When we first started back in 2017, the thought of having to try and record our 17 bat species was was quite scary if you like, but in a way, it was great, because it kind of pushed me into doing it. It's like coming here tonight. I mean, you know, it kind of forces you into doing things. So yeah, so it's meant a lot of travelling around to go to these different routes. Although, back in Surrey, you know, we've got sort of 14 species and you know, you've got to go down to the southwest if you want to record horseshoes, and grey long-eared's and things like that so and come up to back up to the northwest because originally obviously I'm I'm not from Surrey, as most people have probably gathered by now with an accent. I'm from Bolton, so it's nice to get back up to back up north.

Steve Roe: 19:39

And you mentioned all 17 species there, which has been the most challenging species to record.

Unknown: 19:44

Definitely without a shadow of a doubt grey long-eared Yeah, they were my real bogey species. I've had so many disasters, I went to a big maternity at what was supposedly a big maternity roost down in Devon. Only to find the bats weren't there when I got there. I went to Jersey to record the great long is in Jersey and managed to pick the one week in Jersey that year, that had sort of like monsoon rain and gale force winds. And that was just a complete washout. I think at the end of the week, I'd ended up with something like six files of grey long-eared's, is just atrocious. Went through another route in endeavour number to find the brown long-eared's mixed in with the grey long-eared's which didn't work. So yeah, that's been that has been quite challenging. So we've only managed to record at one roost, grey long-eared's. So that's, that's definitely on my site to do list is try and get some more. But for this particular project with BCT, very long is that one of the target species for this project. The main sponsors of the project are Forestry England. And so the focus is more on woodland bats. And so obviously, the Myotis species in particular, are the ones we really targeting.

Steve Roe: 21:12

And finally, if people want to know more about the projects or want to keep up to date, where's the best place they can go and find out a bit more about it?

Unknown: 21:19

Well, at the moment, because the the website is still on the test BCT hadn't really broadcast it. But hopefully, in you know, in a few months time, the website will be sufficiently written if you'd like for that we can actually publish more details about it. But at the moment and BCT are actively trying to recruit sort of volunteers in the membership to actually have a go recording that's at their local lists. So there's some equipment available. So if you contact the BCT and they can give you the details then and they'll send you a set of equipment through the post. And you can do some recording at your your local roost. That's great Martyn cup thanks very much. We better get back and help Dave finish his roost count. A big thanks to Martin for having me along for the evening. And thank you to you for listening to this episode. I hope you've enjoyed it. If you take a look at the show notes, you'll find a link to Martyn's social media and our advice on bat detectors. You'll also find a link to an earlier episode which was also recorded in the Churnet Valley with members of Staffordshire back group who are learning more about Brandt's. This series we are running back chats first ever competition. children's authors Emma Reynolds and Angela Mills have kindly donated prizes. Angela has donated a copy of Bobby the brown long-eared bat signed by both Angela and Chris Packham, and Emma has donated a copy of her newly released book Amara and the bats to enter the competition to win one of these brilliant books. All you have to do is write us a review about this podcast about chat, and the two winners will be picked at random at the end of this series. Not all podcast apps allow you to leave reviews so you can find instructions in the show notes of this episode. And please note that we were only able to post the prizes to addresses in the United Kingdom. The series continues in two weeks time but until then, I'll leave you with the sound of brown spots as they emerge from their cottage in the woodlands of Staffordshire.