hpr-knowledge-base/hpr_transcripts/hpr1894.txt

Episode: 1894
Title: HPR1894: Interview with Davide Zilli and Dr Marianne Sinka of the HumBug Project
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr1894/hpr1894.mp3
Transcribed: 2025-10-18 10:56:46

---

This is HPR Episode 1894 entitled,
Interview with David Millie and Dr. Marion Sinker
on the Hummer Project.
It is posted I can't fall in and in about 61 minutes long.
The summary is mosquito detection
and habitat mapping for improved malaria modeling.
This episode of HPR is brought to you by AnanasThost.com.
Get 15% discount on all shared hosting
with the offer code HPR15.
That's HPR15.
Better web hosting that's honest and fair
at AnanasThost.com.
Hi everybody, my name is Ken Fallon.
You're listening to another episode of HPR15.com.
Today we have an interview for you.
It stems from a blog post that I wrote over on my own website
called Open Source Mosquito Locator.
Basically, what I tried to do in the post
was find a way to locate mosquitoes
because all the things that I've seen up until now
has been ways to destroy mosquitoes.
It's a location that seems very interesting to me.
After that, once you know where they are,
you can identify the species and you can eliminate them
if that's what you want to do.
This stems from my kids having a very bad reaction
to getting bit by mosquitoes.
I don't know if people know,
but the Netherlands is covered in mosquitoes.
Something I didn't know until I was a-married
and be moved over here.
Not that that would have in any way influenced my decision,
but so mosquitoes, though, are a bit of a problem.
There are massive big problems in certain areas
where there is malaria.
And this is with the importation of things
like carotires with little bits of water in it.
Every year, that's beginning to come into certain areas
of Europe as well.
So it's a fairly serious thing.
Mosquitoes are the number one killer of humans
in the world of everything, I believe.
But it could be completely wrong.
So lots of blood posts, lots of comments,
and one comment by Alexander Azali,
who pointed me to humbug.robots.org.ac.uk,
which I'm guessing is an academic institute in Oxford
with something to do with robots.
And I contacted that project,
and that project,
they were very kind to come on and talk to us today.
So if you could both introduce yourselves,
that would be fantastic.
Okay, I'll start.
I'm a researcher on the humbug project.
You, as you said, it's a project at the University of Oxford.
It's a collaboration actually between the Royal Botanic Gardens,
Q and the Information Engineering Department in Oxford.
And yes, we are.
We started about six months ago with this project,
which is funded by a Google Impact Award.
And the target of this project is really to detect
and classify different species of mosquitoes.
But I will let Marianne introduce herself,
and then we can say more about the project.
Hi, I'm Marianne Schinka.
I'm also working on exactly the same project full time.
It's the idea behind the project is that
may be able to distinguish between different species
of the mosquito by the sound of their wing-beak frequency.
So basically the little buzz that you hear,
when you know you're just about to be different at the noise.
And so what the project was,
we were trying to see if we could work out whether we could use that noise,
and we could use other information such as the surrounding vegetation.
And that's why I use involved.
And just mosquitoes.
I'm a mosquito researcher.
I've spent about 10 years looking at malaria vectors
and trying to find out where they are and how they behave.
Because they all can behave quite differently.
And there's a lot more species that people realise.
And we're trying to put all of this together to make a useful sensor
where we can map the role of them.
Also to alert people to the fact that there might be a mosquito
that may be carrying a certain disease in their vicinity.
I think I'll have that back to David to give you more of the technological.
And why did you get into studying mosquitoes in first place?
Well, I've had a background in entomology.
I've always looked at my PhD was looking at insect interactions.
So just looking to see how insect herbivores affect the insect
that live in the soil as an interaction.
So basically you get insects affect a plant in different ways.
And that affects what goes on in the soil.
It can't respond to different subjects.
So if you...
If a caterpillar chooses a leaf,
it won't respond in a different way to if an aphid pierces the leaf
and tries to suck out the leaf juice.
That has a different effect on what goes on in the soil.
So that was my kind of beginning.
And I've always been interested in cells and ecology.
And then I managed to get a postdoc in Oxford looking at mapping
different disease vectors.
And there's a lot of disease.
And there's a lot of mosquitoes that carry them.
And...
Oh, sorry.
Absolutely, no idea.
And that's my one-year-old type of that.
Come back.
Where you go?
That's...
That's LC, little girl.
Good.
Very happy.
Excellent, excellent.
Sorry.
And...
Sorry, that just lost my train of thought a little bit there.
Yes, so basically I've had random insect research.
But I've also been very interested in the medical side of things.
I was able to combine the two looking at insects and looking at disease.
And actually making it have a little bit more importance.
A little bit more focus in why you're actually looking at a particular insect.
And mosquitoes are surprisingly very...
But quite understudied when it comes to vectors, vectoring disease,
which is quite surprising.
I would have thought that...
So given the number of fatalities that it would have been more researched.
Well, what has happened is...
I think people initially considered mosquitoes to be fairly simple organisms.
Which they are.
But there is a kind of little extra bit of complexity that hasn't been factored in.
And I think that most people who started...
Initially started looking at things like malaria, dinghy and stuff.
The focus was on the disease.
So the focus was on the malaria parasite or on the dinghy virus.
And the mosquito was considered just to be the vessel that carried this disease
from one person to another and not much else.
In more recent years, people have realised that this is actually very important
to understand these mosquitoes.
There are...
In the literature, there's people who've spent their entire career just trying to try
and focus on one particular species.
One of the main problems is the fact that it's quite difficult to tell...
Tell some of these species apart.
So you can have two species of mosquito that look absolutely identical.
Perhaps there might be different ridges in the eggs.
But if you were to look really, really carefully, you might be able to distinguish between these two species.
But other than that, they are morphologically identical.
But they behave very differently.
So one would be an extremely dangerous factor of disease.
The other one would have no interest in biting people at all.
And would only, for example, by carols, for example.
So people have done...
They've been working on these things, but they haven't been able to fully identify them.
Until more recently, when we've had the kind of techniques, the DNA techniques,
to be able to pull these apart and say, this is definitely a different species.
This species is definitely behaving differently.
This one is not a danger.
This one is, we need to focus our control efforts on this dangerous one.
And it can all be the other one.
If that makes sense.
It makes 100% sense.
However, here I do need to interject with the question,
well, why not just get rid of all the mosquitoes?
Because, how would you do it?
Yeah, exactly.
Well, presumably, I could put the same question to you.
We identified there's one species that we want to target.
Then how do we get rid of that one as well?
It's the same question.
You could just carpe a bomb the place with DDT
and get rid of all the mosquitoes and all the other...
Yeah, I'm playing devil's advocate a little bit here, but...
You know, a mosquito is a mosquito is a mosquito.
Kill all the mosquitoes and we solve the problem.
Yes, if you kill all the mosquitoes, you all have solved the problem.
But it just is...
I mean, there are a massive campaigns out in...
Wherever there is malaria for people to try and stop the mosquitoes.
So what they do is they try and stop the specific mosquitoes.
Because if you wanted to kill all mosquitoes,
you would have to treat or deal with every water body that there is.
So mosquitoes will be found everywhere in the world,
apart from, I think, in the deepest Arctic and Antarctic, everywhere in the world.
Everywhere that there's water.
Yeah, if there's any place where there's...
There have been times where people have definitely tried to deal with all the mosquitoes.
I remember hearing one story from a colleague and they'd gone out.
And I think it was actually in an area where they were just trying to deal with all the mosquitoes.
And they couldn't...
They couldn't work out where the mosquitoes were coming from.
They had treated all the water bodies.
They're all the obvious water bodies that they could find.
And then eventually they found the mosquitoes were breeding in a golf course in the little holes where the balls go.
So when we're talking about you have to search out every little bit of water.
You have to search out every puddle.
You have to search out every little bit of water that's in every tree hole.
Every little bit of water that's caught in...
If you've got some plant pots in your garden.
Every bit of water that's in those plant pots.
When the rain comes in places in Africa, it just takes moments...
It takes days for the mosquitoes to find...
I'm very glad.
...to emerge and come out.
So in a few days after the rains, you have the mosquitoes coming.
And it's just a huge problem.
And also, obviously, you have the ecological kind of impact.
If you were to go and wipe out every mosquito,
there would be a kind of full-on impact.
But what they're trying to do is they're trying to wipe out the ones that bite the people by putting out fitness.
Over the people that are treated with insecticides or spraying inside people's houses within seconds.
Okay. Have you gone silent for a second?
Okay. So they try to treat it just locally in the house.
And they get mosquito free zones and houses.
And I understand as well the supply of mosquito nets are helping quite a lot.
But what makes you think then that if you can identify a mosquito that you can eradicate it?
Because the idea is...
If you can identify where you've got the mosquitoes that are carrying the disease,
you can focus your efforts and your money in the right place, first instance.
Because, as I said, different mosquitoes will behave in terms of ways and be able to transmit.
And also, the idea...
The other idea behind the sensor is to understand the mosquito better,
to say what is the surrounding vegetation that this mosquito is associated with.
Let's understand a little bit more about the ecology of the mosquito.
What are the very local climatic conditions that are found with this mosquito?
So if they're attracted to...
Another daughter of this, sorry.
How old is she?
She's three.
Wow, we were rising in the middle of it.
Sleepless nights of it.
Oh, yes.
If so, lost train thought.
Yeah, definitely.
Then you can focus your money and you can understand a bit more about the species.
And also the...
Sorry, my second...
Take time.
No.
We can have long pauses of silence if you want,
because we can showcase that out without any problem whatsoever.
So if you need to go away for a few minutes, absolutely no problem.
Okay, thank you.
So, go back to what I was saying.
The current...
Well, one of the issues now is that the begnets and the insect spraying indoor...
Inside houses is wiping out the main important carriers,
the ones that have been doing the biggest transmission of the disease.
But there are other species that are still able to transmit the disease,
but they're the ones that don't come in and bite people indoors.
So they won't be affected by a bed net over someone else's sleeping.
And they don't rest on the inside of a wall in a house.
So they won't be affected if you spray the inside of a house with DDT or whatever or an insect site.
They hang about outside.
So understanding what species are aware, identify them properly.
Identifying when you've got species that are outside,
fighters or indoor fighters means that you know,
do I spend all my money on bed nets?
Or do I spend all my money on trying to find other ways to deal with these mosquitoes?
Is this mosquito, a mosquito that will bite people and cows?
You can spray cows with insecticide.
There's lots of interesting and different methods of trying to control them,
but you really need to know what you're dealing with before you can do it effectively.
Absolutely, cool stuff, cool stuff.
So like, come here as simple as they're attracted to plant X.
So move plant X as far away from people as possible.
Exactly, exactly.
The mosquitoes, they feed on plant nectar to give them energy.
The females feed on blood to allow the eggs to develop.
Males feed on nectar only.
And the females feed on nectar for the energy to fly about.
And that kind of everyday life.
We'll try to find out if there is any association between any specific plants and these mosquitoes.
Okay, fantastic. Thank you very much.
And there is also another side to it, if I can add,
if we understand a bit more about the relationship between the mosquito and the vegetation that they like,
we can hopefully relate to, related to a subtler imagery of the vegetation
model outbreaks of the malaria disease and hopefully predict the outbreaks as well.
Oh, very interesting. Very interesting.
So this is the side the cue gardens are looking to more because they have the background on the plant species
and also the expertise and the GIS and the kind of geographer expertise to relate these two things together
and hopefully together then we can create some models to predict the outbreaks.
Okay. David, if you could explain to us what your background is and how you got into the project.
Sure. My background is actually in computer science,
but I also was always interested in insects and in insects sounds, I guess,
because my PhD was to do with a new forest cicada.
We have run this project in around in a new forest,
the National Park, just a southwest of Southampton.
And the project was trying to detect the presence of this very endangered cicada,
the only species that we have in the UK.
And so what's the cicada?
A cicada is a sort of like a grasshopper or a cricket.
It's very similar, although belonging to a different order is similar to a cricket, let's say.
And it's an insect that it's very common around Europe and the rest of the world,
but it likes warm climates.
So it's not so common in the UK.
In fact, the only species ever found is this species called a new forest cicada
that lives on the south coast.
And it was only ever found in the new forest that's where hence the name.
Oh, cool.
And it's this really, really high pitch sound that is very loud,
but it's so high pitch that most of our, we really struggle,
most of adults really struggle to hear it,
because it's at the edges of our hearing range.
And so we run this citizen science project where we got,
we created a smartphone app to detect the presence of this cicada.
And all the visitors around the new forest,
so lots of visitors downloaded it up and held us and rediscovered this species,
which is going extinct.
And unfortunately we haven't found it yet.
But the project is still running, I guess.
So this is a bit of a long way to say that my background is,
is in computing systems, but with an interesting sound,
especially in insect sounds.
Okay.
Very good.
And on the website there, you have,
so there's three different sections under the project researcher.
Yes.
So I guess you will be the smart acoustic sensor.
Yes.
So my side of the project is on the acoustic sensing, yes.
So what we're doing now is we're developing both a hardware prototype
and some algorithms for recording and detecting the presence of these mosquitoes.
And the first challenge with that is to get a great big data set of mosquito sounds
that we can train algorithms on.
This is algorithms need quite a lot of data to understand what the features are
that we want to extract from the signal.
And once we understand very well from all the different species that we want to classify,
then we can refine the accuracy of the algorithm.
And the algorithm, at the moment, we have a deployed in a,
what was simple version of the algorithm that we have so far
is running in an app for a smartphone.
I think you can see on the website as well.
And what the app is at the moment is not for public use,
it's just for us to record, to go round, to researchers that have mosquito cultures
and record different species and different sounds.
And obviously we're trying to, mosquito sounds very hot with lots of different parameters
in particular the age of the mosquito, the gender, if it's male or female,
the humidity of the environment and if the female is blood fed or not blood fed.
So there are lots of things that can make the sound of the mosquito very.
And we have to get a big data set of all these recordings of all different circumstances
for as many species as possible.
And that's our first step.
I guess this is why Google got involved.
Yes, so I think, so this is the,
this was initially, Google had, I have this impact challenges.
Basically they, they fund projects, charities and just generally projects
that they have an impact on, on something that is, I guess, quite societal change.
And we applied about a year ago, I think maybe a year and a half ago
and we went through all the stages of the interview and in the end we got this funding data
that can help us with this project.
So as you say, this is, sorry, go on ahead.
Oh, sorry, I was just going to jump in.
But the Google impact challenge, what they do is they, they have,
a lot of people put in applications for it and they funded, I think it was 10 projects
in the end and ours was one of those, one of those 10.
And it, it seems that what they do is they look for kind of innovation
and something unusual.
And the, the great value to this is the fact that they will fund projects
that perhaps other funders in,
funders would not consider because they, because there's a risk involved.
And so it's, it's actually a really, really good source of funding for us
at this stage of the project when we're, when we're looking at something that's very, very novel
and that is difficult to implement.
But, you know, we're hoping to achieve.
But it's basically Google is a particular funding source is, is unusual
in the fact that it will fund and it will take a chance on something unusual.
And are there any requirements on them?
I mean, will you be releasing the source code or will you be making this device publicly available
so that a lot of people can use it or how does that work?
Yeah, I think target, I don't know if it's a requirement or not.
I think something we might have to check into.
But definitely as a, as a research project for us,
we will be very happy to make this open source.
But the, but the algorithms will definitely be published if anything.
And the app will be made available in open source.
And more importantly, maybe in my opinion,
is the, the data set that we are collecting so that everyone can, can try to,
to have a go, develop in different algorithms.
I think that's the big thing for me and that we will also be making open.
Oh, fantastic.
If we could just go through the habitat modeling and the vector modeling pages
because the acoustic sensor, I, I kind of get from, from our background here.
It's, but I do want to come back to that.
But I wonder, could you give me an overview of what the habitat mapping,
what the benefit of that is and the vector modeling?
And could I just, with the vector modeling, the, the text there,
I'm just, I'm just rewriting the text that will go on the vector modeling page.
And that will just kind of underline that what we're doing and how,
how we're using occurrence data to be able to what,
or previously how I've done it and how, how I intend to do it,
use occurrence data to and co-variate such as the climate,
such as the plant species, such as the kind of flowers around,
to try and to niche models.
I don't know if you know,
we're basically you, you use lots of occurrence data.
When you say occurrence data, you mean,
when you say occurrence data, what do you mean exactly?
Oh, sorry.
Accurers data.
So basically when you've, we've got a confirmed,
it's a confirmed idea of where,
and where a species is necessarily have to be a mosquito,
but a particular species.
So if you're doing niche modeling,
what you need is you need to know the exact location of a confirmed identification
of that species.
Gotcha.
So you've got all the lat longs for the various different sites.
So you can actually map them.
You can put lots and lots and lots of dots on a map.
And then what you can do at each of those specific locations,
particularly using satellite data,
or other data that spreads across,
or remote sensing data,
or weather data,
or whatever you have that spreads across that area.
And kind of drill down on each of those locations,
and say in this location,
what is the climate?
What is the, what is the land cover?
What is the entity?
What's the normal rainfall, etc, etc, etc.
And then you can say each of these occurrence points that I have,
these are locations.
What is, what's, what's the commonality between all of these points?
What is it that the species is looking for?
So you're looking for...
And then...
And then what do you do?
And then what do you do?
You're looking for this famous plant X,
or a combination of factors to say...
Yeah, you're looking for...
It's niche.
Gotcha.
Yeah.
This mosquito needs all these things in order to survive.
Exactly.
And then what you can then do is you can look
using the habitat surfaces that we have.
So satellite data, for example.
So where else do these specific conditions exist that we have identified
in each of those locations that we found the species?
Where else are...
Do these particular conditions exist?
Then we can say that it's quite likely that you will also find the species in this area,
unless there has been some sort of intervention that has prevented either the species getting there.
So, you know, travel...
Maybe there's a mountain range,
something from where the species are originally kind of evolved.
It's never managed to get to this particular place.
Or, you know, has it been mosquitoes?
Would it have been there if people hadn't already sprayed insecticide?
So what you could do is you could identify what's called a fundamental niche,
which is basically all the locations where the species would exist under ideal conditions.
Yeah, catch it.
And ideally, you're...
If you're getting lucky,
if this was a Sunday night movie, for instance,
you would find some sort of insect or other that is eating the mosquitoes and that other area.
And then you could do all sorts of things.
Well, yeah.
I think of the...
Because what they...
They...
They inspire areas in Asia, for example.
You get a lot of mosquitoes that breed in rice fields.
So what they do in the rice fields is they're really certain...
Certain fish species.
And the fish eat the mosquito army.
And then the...
When they...
They drain the water out of the...
Or the paddy fields to collect up the rice.
They also collect up the fish.
So it's...
It's a really neat system.
Well, you know, getting your rice and getting your fish at the same time.
So this is...
This is real actual science at work here.
Yeah.
Very...
Very, very impressive.
The amount of data that you're going to need to do this thought is going to be phenomenal.
Um...
Yes.
Yes.
Yes, it is.
But I...
I think that they will either reall, disagree, I don't know.
But I feel that once we've managed to get the system set up,
the...
It should be kind of self-perpetuating.
So we should be able to use our sensors to gather our data for us.
And increase our data set.
And use that...
That data set to refine our sensors.
So it should be a kind of continual circle.
So we kind of put out some sensors.
And they are discovering a different sound of a mosquito and a different area.
And the next place we found...
Another different sound, another different area.
And we keep collecting this and keep collecting it and collecting all the data.
And then we can kind of analyze the data.
And see, find, find the associations.
And then once we found the associations and we're saying there might...
There's likely to be a different species in this area than this area.
We can go out and start collecting it.
And also in addition to using the sensor, we...
In a previous project, I went through lots and lots of literature that's available
where people have reported where they've collected different species.
And so what they do is they normally get the site and they normally give the coordinates.
And I've extracted all that information and mapped it and created some of these niche models.
So we are kind of trying to refine these niche models.
And then what we can do is we can try and associate sensor data with the niche model outputs.
And see if we can find a pattern.
And then we can see if we are able to distinguish between this.
And if we are able to pick up different species and different conditions that each species is occurring in.
So you have other sources of information.
So like satellite imagery, where is that coming from?
Well, there's a huge amount of open source satellite imagery that is just available to anyone and everyone.
Quite a lot of the NASA satellites produce lots of these kinds of grids.
And look at the satellites continuously circling and collecting all this information.
And all this information is just there to be downloaded.
It just takes a bit of a bit of processing to just.
I say that it takes a bit of processing.
I know it takes a lot of processing for people who who have to do it.
But I always get the end results.
So it's easy to say a bit of processing.
But yeah, it takes a lot of computing power to type to download all this stuff and clean it up a little bit.
Try and remove because obviously the satellite can't see through cloud.
So you have to kind of work out what's going underneath on underneath the clouds.
And reinterpret it and then you have to kind of try and put the clouds back in again.
Because clouds can be an important climatic factor that we need to consider.
So yes, it's actually a lot of processing involved.
But all of this stuff is there is freely available.
And there's lots of software that to help people.
But within in Oxford, there are a lot of people working on these things.
I think.
And most of the stuff is freely available, even post processing.
And one thing that the Marianne just touched on, I think is very interesting.
The fact that the satellite images are stripped of the cloud coverage because they try to find ways to remove the cloud coverage.
Because that's something that is in peace, you know, seeing what you would like to see.
But actually at the end, they realize that in many ways the amount of cloud is important as a climatic factor.
So it's the information that it has to be put in again to understand what kind of climate is in that area in that part of the year.
So I think it's a fascinating problem that people try to find is a challenge, you know, a problem to have this cloud coverage on the other images.
And then actually that's something that when you manage to remove it somehow, then you have to consider it again.
Because it's an important factor actually.
Excellent. That's a fascinating stuff.
So tell us about the smart acoustic sensor, which is the thing that I personally am very interested in.
Yeah, so Marianne said, I'm going to disagree or agree with her.
I fully agree that that's what we're trying to do.
So to get the senses to do the work for us, to collect the data in a very distributed way.
And the challenge is that the mosquitoes are incredibly quiet as you know.
You might not sound as same quiet when it's annoying you at night.
But in general, obviously, you can only hear a mosquito when he's a few centimeters away from you, even in a quiet room.
So what we have so far is a small prototype sensor.
And you can see we don't have a picture of it on the website.
But if you go to soundtrap.io, that's the soundtrap.io, that's the project where we develop these sensors.
And these are low cost, low power acoustic sensors, acoustic loggers.
So they have a very cheap microphone and a cheap and bad microprocessor.
And they record and log to an SD card.
So that's one side of it.
And that's developed by Professor Alex Rodgers here at University of Oxford as well.
Just give me one second. I'll give people the specs.
It's a 67 megahertz LPC 1768 processor.
We have recordings to SD card.
48,000 samples per second power by three AAA batteries.
It's got a microphone measures 50 by 38 by 12 millimeters and configured over a USB interface on board real-time clock.
Cool.
Fantastic.
And yeah, so this is not as low power as it goes.
Obviously, you will know that we could make it a lot more low power.
But when we are there is a PhD student at Southampton University,
where both me and Alex Rodgers were before,
that is looking at an ex-generation of this of this logger.
And these loggers are actually not used just for mosquitoes, but we're using this PhD student,
just using them for logging other events, other acoustic events like illegal logging.
And this is the logging in a different sense, the different meaning of the words.
Illegal logging in the forest in the Belize.
So there are other scenarios where these same sensors could be used.
And the challenge, as I said, is the how far can we hear a mosquito,
how far from the device does it have to be, from the microphone.
And I mean, there's no definitive answer because it depends on the background.
No, it's more than anything and different things.
But you only have a few centimeters to catch the sound.
But the good thing is that mosquitoes actually use sound to communicate.
So it's obviously a very good way of detecting them.
You may know that they use sound for mating purposes.
So they choose a partner with the sound.
They try to match in frequency.
Yeah, the mosquito of the opposite gender.
And sometimes, if they can't match in frequency because their size is too different.
So the frequency is too different.
Then they will match other common harmonic, which is something I find incredibly sweet.
And they will, for example, match at a third harmonic of one and the second harmonic of the other.
So something that is actually quite refined in the system is the organ that is responsible for their hearing.
It's called the Johnson organ.
And there are researchers.
Some of the researchers have been helping us a lot, like Professor Gay Gibson and University of Greenwich,
that has been studying mosquitoes and their sound for a long time.
What do you...
Sorry to interrupt.
If you could then emit the sounds of, you know...
You could use that sound emission then as a trap or a lure to get mosquitoes to the device.
Absolutely.
It's been shown that mosquitoes actually respond also to a pure tone of the opposite gender,
of the same frequency of the opposite gender.
So luring them would be an absolutely great...
I mean luring them with sound, I mean it would be a great way.
There are other ways to lure mosquitoes.
They are attracted to light.
As you probably know, they are attracted to CO2 because they see that as an animal breathing.
And so a combination of these things is what normally is used as a trap for mosquitoes.
So many of the mosquito traps that are commercially available would have a light
and a fan blowing on some dry ice to generate some CO2.
We've got some of our senses deployed in close to these traps,
in walling for close to Oxford, with the help of the Centre for Ecology and Hydrology.
And the problem for us is that these fans obviously make a noise.
The noise that is quite live compared to the mosquito.
So yes, that's one of the other challenges, exactly.
So one of the...
What we hope is one of the previous Google impact challenges,
I think it was the year before ours, was a group of people who were putting together the solar lamps
that they hope to distribute across Africa.
So to allow people that don't have access to electricity,
to be able to read or continue their life after dark.
As we kind of considered to be quite easy to do.
And we were one of the guys in our Humbug group was speaking to one of their guys.
And we are hoping that we might potentially be able to put one of our senses on one of their lights.
And the idea being that we can, by combining these two, we have light.
They'll be using the light. We'll have someone sitting there reading or what are they doing,
producing carbon dioxide.
They are basically going to be a massive target for any mosquitoes in the area.
And therefore we were able to kind of collect this information.
And at the same time, you know, provide them with information saying,
yes, there are dangerous mosquitoes in this area.
You need to protect yourself.
Most people in most places in South Africa know that there are dangerous mosquitoes.
But, you know, you could probably change the color of the light or something.
Okay, yes, they were obviously a very good idea.
I'm turning on the light service.
Problem is people get carried away. It's the identification that's the hard problem.
It's just flashing right in the light.
Okay, sorry, got way later.
The device though, why build a specific device?
Is it not integrated into a smartphone or something like Raspberry Pi or something even cheaper?
Yes, absolutely.
I mean, two things.
One is that we also are integrated into a smartphone.
The other side of it is that we have this app that I was saying that can record
and analyze a bit of the sound with the initial version of the algorithm
and tell us if there is a mosquito round or not.
That's on the smartphone itself, yes.
Okay.
And that's the kind of picture that you see on the acoustic sensor on the page of the website.
And that is actually the cheapest option because we are using this Alcatel Pop C1 smartphones.
The Android 4.2 and we can buy them from an e-shop for £9.99.
That is also very difficult to beat when you develop a custom hardware.
It's got a screen, a 3G Wi-Fi, a fairly powerful CPU.
Everything that you could possibly, obviously a microphone,
if you are aware to put all these components together yourself, then it will be very difficult to beat it.
Just a personal interest.
Sure, this is an e-shop, an absolutely high-street shop, nothing special.
We don't have any particular deal with Alcatel.
I suppose they are heavily subsidised by EEE.
But still you can get them with a SIM card for only £10.
Wow, cool.
Sorry, yes, I lost my print of thought as well.
Yes, these devices are very good.
The acoustic logos, we can manufacture them for about £25 at cost with all the components.
It's probably cheaper than in Arduino or something similar.
Erasbury Pie would be good, but it will require a bit more power, I suppose.
We are hoping to get this down to a very low power.
I think...
Oh, sorry.
Go for it.
I think we all know that you may have already said this.
The smartphones in themselves are just this ideal, tiny, already-made bit of kit that has everything you need and it's got the GPS in it.
It's got the recording system in it.
It's got the ability to do the analysis already in there.
Whereas things like the Raspberry Pi, you have to add on all these peripherals.
You keep adding on, and every bit you add on adds more cost, every little bit you eat.
All of these things have been trialled by someone else on a different project and trying all the different ways to do various different monitoring.
It all came down to, let's just use something that's already there, like a tiny little computer that someone has already built and they can build it far cheaper than anything that we can do.
That is your common smartphone.
Okay, why specifically than the soundfap that I owe a project if you can guess smartphones were so cheap?
Well, I guess power is the main reason behind that.
There we can make these devices, there are low power enough that it can last for months on three AA batteries and hopefully with a very small solar panel they could run indefinitely even.
Which is not impossible with the phones either.
One of the people we collaborate with is Russ Laiberi, who also happens to be a Marianne's partner.
He is in Oxford, doing research and energy and he is the one who introduced us to this fantastic Alcatel phones, but he also run a project over in Oxford, powering one of these phones of a solar panel that you can buy on Amazon.
On a rooftop of Oxford buildings, so on a winter, so it's not impossible to do power these devices for a long time as well.
He ran it for six months just off that solar panel.
If you can get phones for 10 euros, 10 pounds, then you're winning.
Different devices for different situations.
Obviously, if we can get lots of people to download an app and on their own phones, that's fantastic.
But sometimes you do need to have something a bit more bespoke for what you need.
That's why we're developing the sensors, so we've got two avenues to try.
I think David will disagree with the bespoke one, even though it's not going to be more costly, you can tweak it a little bit more in a better microphone.
You can actually make it the ideal, and then we can see what we can replicate with the phones.
I guess it's just trying to do two different ways to do it and see what works best.
The one thing about the ARM embedded device, how do you know where that physically is and how do you collect the data from that?
Sorry, sorry, you.
Yeah, your connection just went a bit zylon there on us.
But with the ARM embedded device, the sound trap that I owe, it doesn't appear to have a GPS module and it doesn't appear to have an internet connection.
So you would physically, a researcher would need to physically drop it somewhere and make note of where it is and then go back and collect the data from time to time.
That's right, it's right, so there was a need for the invitation agent of that device.
Oh, David, are you sure I'm bored?
Can you just connect or reconnect again?
Sure.
Is it better now?
Yeah, much better, much better.
Okay, great.
I'm not sure if I saved the recording actually, so let me just start it again just in case.
Cool.
So yeah, that is a limitation of the device, unfortunately, but we could add a GPS or a GSM module.
Maybe I'm in a data connection would be very costly and possibly also not not available in the places where we would like to deploy it, both in the forest and beliefs or in a
around the places where we would like to monitor malaria.
I don't know if we will get a good internet connection, certainly or most likely not a Wi-Fi anyway.
But with a small GSM module as well, that could be done via text message or something.
Unfortunately, obviously all these components add to the cost and enter the energy cost.
But I think here the approach you have is like we've got one device that's would be there, but could be put by a park ranger somewhere who will go around and then collect the data.
And then the other devices are for the people who already have smartphones or this is a smartphone.
So yeah, that sounds like an idea plan.
Have you ever considered like making the requirements available on the internet so that like I at home with my electronic devices here, I do have a Raspberry Pi.
I do have an internet connection and I could then start adding us adding that sort of information up to a public site.
There's something we haven't considered yet, but I don't see why not actually it sounds like a very good idea.
That's something we should think about it, yes.
The sound trap to IO is also aiming to be open hardware.
I mean, as I said, Alex Rogers is the person who develops it.
So I don't want to speak for him until he's done it.
But I think the final idea is to make that open as well so that people can replicate it easily at home.
I might actually contact him and see if we can get him on for two years or?
Yes, absolutely. He will be happier.
Cool. Excellent. So how long do you have the funding for?
Is it for years or for more longer term now?
So the funding is for three years and as it is now, we've got just over two years left.
And that's the initial funding is now unlikely that we will find funding to continue this work for the longer term.
It seems like an absolutely excellent project and something that obviously I agree that we need to be doing.
Thank you.
Is there, is there.
If people are interested in this, obviously we go, you can go to the website.
Is there do you update the social network with your project?
Where do you keep updates and are there we wait for the paper to come out?
Yes, there's something we should also be a bit more active about at the moment.
We have a Twitter handle but we haven't actually used it yet.
And so I think we will update our website with a news section and we can keep our people up to date and people can subscribe to it.
But yes, unfortunately at the moment, we have to say not yet.
No.
It's actually very important for a project like this that we do try and get a lot of people interested in it because
I think once we do have everything set up as we want, we manage to get our algorithms available for an app for a smartphone in a way that we're willing to kind of try and distribute.
Obviously the more people we can distribute it, the more people who we can get to have a go with it, the better and the more our data set will be improved and the more the whole outcome will be improved.
So yeah, I think we do need to be a bit better on our website.
Absolutely.
No, so I think when people think of mosquitoes, right, they're thinking of
poor areas in Africa and swamped regions of the world.
But it's not just limited to that.
I mean, obviously here in the Netherlands, we have mosquitoes, but you say you mentioned that you have this going in the UK.
Are there mosquitoes in the UK as well?
Oh yeah, yeah, there's plenty of mosquitoes in the UK.
And in fact, there's even mosquitoes that have the capacity to transmit malaria in the UK.
And malaria used to exist in the UK.
It was a white, but basically malaria is a disease of kind of poverty.
When a country is able to kind of develop well enough to deal with it, just simple things like windows, for example.
That will kind of stop the transmission.
And yeah, in down in Norfolk places like that, you know, the very kind of boggy places,
Norfolk broads and so on.
Yes, that malaria was was around, resisted and was transmitting.
I think I think I would need to just double check it.
I think it was transmitted in the UK by a parasite that's now extinct.
But yeah, if we were to get enough people coming over with malaria into this country, into the right area,
you could get someone likely.
And as it's linked, the device is linked to just mosquitoes.
I can identify other species are we or are you just particularly targeting mosquitoes?
At the moment, we are just targeting mosquitoes.
There will be a tune for mosquitoes in particular.
I don't know if the audience is interested in the details, but I think at the moment,
the very simple thing that we are doing is to extract two frequency bands from the spectrum,
which is a very efficient way of extracting a single bin of the FT.
And we're taking these two frequencies.
One is at the central frequency of the mosquito sound.
And one is at a frequency where there is no mosquito.
And hopefully as little background noise as possible.
There is no issue of these two frequencies, frequency bands, so that it's this ratio is high in the presence of the mosquito.
And it's kind of around one in the presence of white noise that covers the entire spectrum.
And this should be very low in every other case.
And so this is something very simple that it will be improved over time when we have a lot more data.
It's not specific to mosquito, although it's tuned to mosquito, if you see what I mean.
So the frequency is a way different for mosquito frequencies.
But if you wanted to look for some other sound, that would work as well.
Oh, fantastic.
So I could theoretically get a microphone into my thingy.
Is this a physical filter or is this a software filter?
It's a software filter.
Okay, cool.
I really would love to be able to download this and compile this software myself and just use my own laptop to start getting this information.
Okay, sure.
Yes, absolutely.
Right, is there anything else that I've missed or do you think that you want to talk about?
There is only one thing that I noticed on your blog post and on the answers that you're talking about locating a mosquito in a room, right?
And I think that's also something very interesting.
It's something that we've been contacted about recently as well.
It would be interesting if anyone had any input on how to do that.
I think it's not trivial.
In theory, you probably only need three microphones.
But in practice, sorry, this is the technique with sound, obviously.
There might be other ways.
I mean, I saw a lot of the comments talk about cameras and yes, obviously,
would need a quite a high resolution camera, I imagine.
But to do it with sound, obviously, it would be possible as well.
But what people do as far as I know to locate sound is to have a microphone array.
So to have a large number of microphones.
But yes, if people want to discuss that or have any ideas of how to do that acoustically,
then I think it would be very interesting to explore as well.
I think a good start would be even if you had one microphone in the room and you could say,
there is a mosquito in this room somewhere or there has been a mosquito in this room in the last ten minutes.
Yes, absolutely.
That's for sure something.
You could make a little box with a little red light on it going, there's a mosquito in the room.
Okay, I will find this mosquito and I will kill this mosquito.
Yes, absolutely.
Okay, absolutely fantastic.
Very, very interesting.
And I know Dave and all the other people who are listening to this are going,
you're asking more questions.
So can we follow up again with emails or how can...
Absolutely, we'd be very happy to hear from the Twitter account.
Yes, so I will have to find the Twitter handle because it's so bad that I've actually
forgotten what username is.
But it's certainly not just handbag because that was taken already.
I think it's handbag mosquito that I will find that out for you.
Thank you.
Otherwise, please, please, please, please, drop us an email from my website.
And I would be happy to get in touch with you.
Well, I'm absolutely astounded how far you've come with this
and I really personally would love to start messing with this as well or getting the app.
It's really what you're doing is actually really cool science A.
But the end goal of your project is absolutely fantastic as well.
And the fact that you don't want to eradicate all mosquitoes from the panches is also kind of nice.
Excellent, anything else there?
I think we're told from us.
Can I ask just a question that's of interest to me as a as a father as well?
Do you work from home a lot or how does that how just having a small family?
How are you how are you getting a work like balance?
Is that going okay for you or is that something you don't want to talk about?
Sometimes it is a bit tricky.
I think both of my my goals have been to what Elsie has been to.
Every home bug meeting that we've had.
Luckily we work with a group of people who are incredibly generous and happy to have me drag Elsie along to most of the meetings.
I do work from home a lot as well.
I do what I have to do is I have to work some very long days.
That's how I can fit my hours in and then do childcare for two days a week.
It's quite tricky sometimes but I said with the kind of generosity of the other people I work with and they're very
flexibility that we're allowed at the university as well.
It makes life a lot easier than I know for quite a lot of other people.
I'm very lucky.
Actually I'm really glad that that is the way and I don't think you.
I would like to live in a world where you don't consider yourself lucky that that is the norm.
I think you know having kids around is just kind of normal.
In the way things are now it's considered odd but I personally think it's normal and yeah that's life.
Very nice to live in a society where you know that that can happen.
So cool good stuff.
I just found the Twitter handle.
If you want to come back moz.
So it's a handbag moz M O W Z.
So in the future I think I'll tweet today our first tweet so that we can start our feed.
Excellent. We'll send some tweets your way as well.
Anything else guys that I missed?
I don't think so.
I think it's very good.
And listen if you ever want to do a follow up show yourself you can just record it and upload it.
We're absolutely more than more than happy to take we are ourselves an open source open community.
So if you are anybody else want to contribute a show on a particular topic feel free to do so.
We have a mailing list as well.
If you want to fire up if you're if you're looking for people to trial anything.
We have fairly technical people who listen to us and they're well used to submitting bug reports and stuff like that.
So I'm sure lots of people will be very interested to help.
That is really good to hear actually.
In fact actually that is one point that perhaps I should make.
But regarding open source and I think that the open source data open source work.
And that's that's very much.
A very important factor of work that are the group that I've been working prior to this and with the homework.
And I think actually in science as a whole.
A lot of the journals that used to be used have to pay to access and now open access.
And that's that's a movement that's very very.
Sorry.
I'm actually kind of trying to wrangle both both girls now so it's not not so.
Not so easy.
I just wanted to just make the point that yes that it's really good that that's science itself and and things like your radio show all this is open access.
I just love it.
I just love the fact that things are now accessible to everyone apart from you.
As opposed to elite academic or I think everyone can now have a look at all this kind of stuff.
It's great to have.
Yeah.
Absolutely.
What I want to do.
I can look at your point because right now this is the part that I was stuck on the.
I was trying to get hold of some academic information as to the frequencies at which mosquitoes operate.
And I've met several requests for academic papers from people who have been researching this and nobody has replied back to me.
And then all of a sudden find out about your project and boom you have you have already the problem solved and you know we could we wish.
Well, I hope so like that we can at least you've got an approach that seems that seems logical and we can try out and if it doesn't work.
Well, that's half the point that's you know that in itself is a is is a benefit for the next person who comes along with.
Well, let's not try that field of research because we know it doesn't work.
So let's move on to something else.
I think it's I think it's also what you're doing.
So cool.
Yes, I agree.
I mean, given that I could be spayed by by taxpayers money.
I think it's is even too late that this is this is happening.
The things are moving towards open source and there's still a lot of work to do to open more of the academic papers and academic journals and.
Yeah, to to to free access and and hopefully we will we will get there one day.
Excellent. Okay, with that I'll wrap it up and I'll say as everybody else who's listening to this remember trying contribute some shows.
And if you come across an interesting project just get on the phone contact them and you never know you might be talking to two.
Two very nice people on the airwaves and so I'd like to thank you both again for coming on and tell our listeners to tune in tomorrow for another exciting episode of hacker public radio.
You've been listening to Hacker Public Radio at Hacker Public Radio dot org.
We are a community podcast network that releases shows every weekday Monday through Friday.
Today's show like all our shows was contributed by an HBR listener like yourself.
If you ever thought of recording a podcast then click on our contributing to find out how easy it really is.
Hacker Public Radio was founded by the digital dog pound and the infonomicom computer club and is part of the binary revolution at binwave.com.
If you have comments on today's show please email the host directly leave a comment on the website or record a follow-up episode yourself.
Unless otherwise status today's show is released on the creative comments, attribution, share a life, 3.0 license.