• Welcome to BirdForum, the internet's largest birding community with thousands of members from all over the world. The forums are dedicated to wild birds, birding, binoculars and equipment and all that goes with it.

    Please register for an account to take part in the discussions in the forum, post your pictures in the gallery and more.
Feel the intensity, not your equipment. Maximum image quality. Minimum weight. The new ZEISS SFL, up to 30% less weight than comparable competitors.

24/7 Raspberry Pi/birdNET monitoring (1 Viewer)

Max_

New member
Germany
I just came across a product from a US start-up called Haikubox (HAIKUBOX).

Apparently, they offer a hardware device (a “box”) which you can place in your yard (or wherever) and which is constantly recording bird sounds around the box (seemingly from quite some distance), identifying them and sending the information to you via an App. The identification also seems to work based on BirdNet. To function a WLAN connection is necessary.
Generally, that sounds comparable to PUC to me. However, I am not quite sure if PUC will offer the same features (such as a bird alert, information sharing from "your" box via the app, ...).

Does anyone of you already have a Haikubox and has any experience to share?
 

LittleBitOfBreadNoCheese

Well-known member
Scotland
I just came across a product from a US start-up called Haikubox (HAIKUBOX).

Apparently, they offer a hardware device (a “box”) which you can place in your yard (or wherever) and which is constantly recording bird sounds around the box (seemingly from quite some distance), identifying them and sending the information to you via an App. The identification also seems to work based on BirdNet. To function a WLAN connection is necessary.
Generally, that sounds comparable to PUC to me. However, I am not quite sure if PUC will offer the same features (such as a bird alert, information sharing from "your" box via the app, ...).

Does anyone of you already have a Haikubox and has any experience to share?
The product was mentioned by the BirdWeather guy when he announced his project, see this thread


"after speaking with Stefan and understanding their (Cornell's) relationship with Haikubox, I have decided to pursue this as a pure passion project"
 

jurek

Well-known member
I would love to have such a thing but offline algorithm. The software can simply have periodic updates of the algorithm.

First, I am concerned about privacy, because it is downloading to somebodys database what I am talking and hearing. These same approach killed "Google Glass" - people objected to being filmed without their consent. Second, I will use it most often in the backcountry with little internet coverage or abroad, where I would get improbable roaming cost. The result “you passed a rare bird calling 2 hours ago” would p*ss anybody off.

The mixed results of AI identification of bird calls so far are another topic.
 

LittleBitOfBreadNoCheese

Well-known member
Scotland
I would love to have such a thing but offline algorithm. The software can simply have periodic updates of the algorithm.

First, I am concerned about privacy, because it is downloading to somebodys database what I am talking and hearing. These same approach killed "Google Glass" - people objected to being filmed without their consent. Second, I will use it most often in the backcountry with little internet coverage or abroad, where I would get improbable roaming cost. The result “you passed a rare bird calling 2 hours ago” would p*ss anybody off.

The mixed results of AI identification of bird calls so far are another topic.
I assume so far most people have them in their backgardens in which case they are less intrusive than say a doorbell camera which are getting quite popular where I live. But if you have say neighbours parking adjacent to the location and you strapped it to a tree and they find it, that might cause an issue. I have people talking on mobile phones outside my window most days. I'm still trying to figure out where I could place a device.
 

Jon.Bryant

Well-known member
Interesting point about privacy. When I listened in to one of the BirdWeather live streams from Florida, I could hear voices - did the people know their conversation was (albeit quite quietly and indistinctly) being broadcast to the world?

I use passive recorders and from time to have caught conversations of passers by. However, I have always deleted the dialogue, selecting the section of conversation using the sonogram. I have never listened to more than a snippet, just to confirm it is indeed a voice that was accidentally captured. To be honest I have more interesting thing to do than listen to conversations of random strangers, and not enough time to try and clean up recordings by listening and manually deleting sections of human vocals.

I am a fan of passive recorders and think the British Trust for Ornithology should update their monitoring methods to allow recorders to be used in conjunction with other methods. Passive recorders could be especially useful for surveys for owls, woodcock, nightingales - which have all been the subject of recent BTO surveys. Perhaps they could also be used for some remote squares in the Breeding Bird Survey, or for surveys of species that vocalise very early, such as Goshawk. I recall however, that in a passive recorder review, BTO stated that passive recorders should be used where you have the landowners permission. How many amateur sound recordist seek landlords permission before they do any form of sound recording? I certainly don’t. I suppose it doesn’t matter whether the recording device is hidden or visible, whether it is autonomous or manually operated - some people may object to being recorded either by a hidden box or a person waving a directional mic about. As a parallel example, I recently took a photo of my wife, who was taking a photo of a bird. My photo had the bird in the foreground and my wife in the background. Unfortunately my photo also caught another birdwatcher in the background, who took exception. He was perhaps legally correct that I hadn’t asked his permission, although I think from his outburst, I was more troubled that he was about to commit the more serous offence of affray!

Are concerns over the individuals rights of privacy taking things too far? I think this could particularly be the case when considering the occasional and accidental recording of a fragment of a random strangers conversation - which we have no intention of listening to in any case. If I set up a listening device in my garden but not too distant from my neighbours window, it would be a different matter - I then think I would indeed have acted unreasonable and disrespected the rights to privacy.
 

Jon.Bryant

Well-known member
Answering my own question. In the U.K., apparently, only need to ask permission to photograph, video someone or record their voice, if they have an ‘expectation of privacy’. This expectation is generally reasonable in their own private space, is not unreasonable to assume in public spaces. Therefore, if you accidentally record the mobile telephone conversation of a person stood on the pavement outside you house, someone on a public common, on a public footpath etc. then you are not infringing the individuals expectation of privacy.

I suppose if you stray onto private land and record the land owner, then you could be potentially infringing their expectation to privacy. Best to not stray and instead ask the landowners permission to enter onto their land and to make recordings.

Whether you neighbour can have a reasonable expectation to privacy when having a loud conversation in their garden, which could easily be overheard by neighbours, I am not sure.
 

LittleBitOfBreadNoCheese

Well-known member
Scotland
Interesting point about privacy. When I listened in to one of the BirdWeather live streams from Florida, I could hear voices - did the people know their conversation was (albeit quite quietly and indistinctly) being broadcast to the world?

I use passive recorders and from time to have caught conversations of passers by. However, I have always deleted the dialogue, selecting the section of conversation using the sonogram. I have never listened to more than a snippet, just to confirm it is indeed a voice that was accidentally captured. To be honest I have more interesting thing to do than listen to conversations of random strangers, and not enough time to try and clean up recordings by listening and manually deleting sections of human vocals.

I am a fan of passive recorders and think the British Trust for Ornithology should update their monitoring methods to allow recorders to be used in conjunction with other methods. Passive recorders could be especially useful for surveys for owls, woodcock, nightingales - which have all been the subject of recent BTO surveys. Perhaps they could also be used for some remote squares in the Breeding Bird Survey, or for surveys of species that vocalise very early, such as Goshawk. I recall however, that in a passive recorder review, BTO stated that passive recorders should be used where you have the landowners permission. How many amateur sound recordist seek landlords permission before they do any form of sound recording? I certainly don’t. I suppose it doesn’t matter whether the recording device is hidden or visible, whether it is autonomous or manually operated - some people may object to being recorded either by a hidden box or a person waving a directional mic about. As a parallel example, I recently took a photo of my wife, who was taking a photo of a bird. My photo had the bird in the foreground and my wife in the background. Unfortunately my photo also caught another birdwatcher in the background, who took exception. He was perhaps legally correct that I hadn’t asked his permission, although I think from his outburst, I was more troubled that he was about to commit the more serous offence of affray!

Are concerns over the individuals rights of privacy taking things too far? I think this could particularly be the case when considering the occasional and accidental recording of a fragment of a random strangers conversation - which we have no intention of listening to in any case. If I set up a listening device in my garden but not too distant from my neighbours window, it would be a different matter - I then think I would indeed have acted unreasonable and disrespected the rights to privacy.
Do they turn down records based on passive audio recording? One can see their point of view with respect to reliability - for example Goshawk is a not infrequent false positive in the birdNET app for me. And birdNET-Analyzer has been reporting Woodcock for me, admittedly at low score.
 

Jon.Bryant

Well-known member
Do they turn down records based on passive audio recording? One can see their point of view with respect to reliability - for example Goshawk is a not infrequent false positive in the birdNET app for me. And birdNET-Analyzer has been reporting Woodcock for me, admittedly at low score.
Well BTO don't currently turn down or accept passive audio, as there is no way at present to incorporate passive recordings into BTO survey work (as far as I know). I think their survey methodologies are currently silent on the use of passive recorders, and provide no guidance on how the technology could be incorporated into a prescribed survey technique.

I asked a local co-ordinator for the BTO Nightingale survey if I could assist with my passive recorders, but was turned down as it was thought that the recorders would not comply with the survey method. Basically you had to stand and listen for I think 1 hour at a certain time of night and at a certain time of year. I am not really sure what the difference would be if you listened to a random (but compliant) 1 hour segment from the recording, as it would be pretty similar to having stood at the same location and listened live. Passive recorder generally do not captures more than the human ear car hear - you are not using a parabola or in fact a directional mic. I can see however, that if you trolled through every hour of recordings to detect a Nightingale, then you would be generating bias as you would in affect be surveying one site more intensely than the rest.

I wouldn't suggest that auto detection was used as the basis for a survey. I was more thinking that you could use passive recorders to cover more locations and at antisocial/inconvenient times - much easier to listen to a recording in the day time, than to be out listening for Goshawks at 4:00am in the morning! I mention Goshawks as my friends found a numerous breeding pairs near their home in France, by listening for calling birds at dawn. They monitored the nests, so no false-positives in their own detection skills!

I think there are problems with auto detection at various levels. Firstly I think the pattern matching is best with complete song phrases, but generally less good with short sounds and particularly calls. There is then the fact that sounds can be surprisingly similar, so in some cases I think systems use location as the only real differentiator - you may therefore not be able to auto ID a vagrant, if it sounds substantial similar to a regional species. I recollect setting up a recorder in Cornwall and passing the recording though BirdNet, which detected 5 different American vagrants! There were no other American vagrants in the UK at the time, and think that the analysis software had simply used the American default, as it didn't know the true recording location. Either that or I was very fortunate in picking the location! But who knows it was in Cornwall in October, so perhaps one of the sounds could have really been the species identified! Finally, I am not sure how sensitive detection really can be - by this I mean that software like Kaleidoscope, needs a certain differential in dB between background noise and the call to 'detect' a possible bird sound. I presume that this is also the case for BirdNet etc. When I look at sonograms, I can see (and then hear) faint calls that seem to escape detection by Kaleidoscope. I have on several occasions run BirdNet (or perhaps it was Merlin) to ID a bird, only to find nothing was detected. I think in general our ears are better at distinguishing a sound than currently available detection programs.
 

LittleBitOfBreadNoCheese

Well-known member
Scotland
Yes I've just been playing around with customizing the species list in the analyzer tool, there is a quick way to do it by just passing long and lat on the command line and optionally week of the year too. It helps chuck out a lot of the real fluff but also risks losing something rare and genuine too. I think the seasonal option might be a step too far, will have to experiment further. My latest command is

BirdNET-Analyzer.exe --i example\myrecorder --o example\myrecorder --threads 3 --lat 5... --lon -3.. --week 4

But on your point about the Nightingales, I would agree with your perspective.
 

Lerxst

Well-known member
So I had a Pi sitting around and having come across this thread last week, I decided to give it a go. Just to get up and running I used the only USB mic in the house -- a cheap desktop unit -- and the system still seems to function okay despite it being the weak link.

I did a short write-up with photos, located here: Automated Bird Audio Recognition With A Raspberry Pi

My next task is improving the front end of this thing, starting with a decent omnidirectional mic.
 

LittleBitOfBreadNoCheese

Well-known member
Scotland
So I had a Pi sitting around and having come across this thread last week, I decided to give it a go. Just to get up and running I used the only USB mic in the house -- a cheap desktop unit -- and the system still seems to function okay despite it being the weak link.

I did a short write-up with photos, located here: Automated Bird Audio Recognition With A Raspberry Pi

My next task is improving the front end of this thing, starting with a decent omnidirectional mic.
Great write up, thanks. I think the chance of spotting something not previously observed is one of the big pluses. How weather proof is your setup do you think? Do you get a lot of wind and rain in your area?
 

Lerxst

Well-known member
Great write up, thanks. I think the chance of spotting something not previously observed is one of the big pluses. How weather proof is your setup do you think? Do you get a lot of wind and rain in your area?

The box appears like it will hold up pretty well around here. It is specifially made for electrical components and connections out of doors. I'd like to add another small box just for the power adapter/extension cord, too. It is the microphone that has me concerned. I want to put a good one on it but I don't want the elements to destroy it. I was thinking of having a small inclined roof that the the microphone can sit under and not get buried in snow. Wind and rain are common in the spring and summer. My yard is small, and the distance from shelter under our deck out to the feeders and trees is only a few meters, so it is fine there, but I want a very robust design anyway. It would potentially be interesting to take it on road trips, for the occasisons where I would have somewhere safe to set it out overnight, etc.
 

Users who are viewing this thread

Top