How to make very long recordings (of bees)

Question

I've been thinking for a while now about trying to make some really long recordings.

I got involved in a film project about honeybees some time ago, and as part of it I would really like to record the sounds and vibrations inside a bee hive. I'm also interested in the changes in these sounds over long periods of time - the colonies develop over the seasons from nearly dormant in the winter to tens of thousands of bees in the summer. This got me thinking about how amazing it would be to make a continuous recording of a whole year cycle of a bee colony. Imagine the spectrogram.

Now the beekeepers in question are not necessarily interested in opening up the hives to extract honey, so there's potential for the recording to have no human intervention. But the question of how to do it is a difficult one. I'd really like to have at least two mics in the hive, and I'd like to record uncompressed. A quick calculation of Stereo 48k 24bit for one year is 8.2 terabytes. In a way, I'd rather not rely on a computer for this but can't think of any other way. It's possible to build a RAID array of that size (with redundancy) and you could have a UPS system in case of power cuts, but I still don't like the idea of trusting a regular computer to this. The other question is some kind of recording software to write the files, it would need to automatically write and name files to fit them into the maximum file size for the OS.

Other questions are where and how to mount microphones? I'm thinking miniature omni lav mics, but the hive might need to be designed specially for these as the bees fill the entire hive with comb and encase foreign objects in wax! Another option is contact mics on the hive itself - that's actually quite an interesting route as the bees communicate in vibration across the comb itself.

I could visit the site every month to check it's still running, but it would need to be self sufficient between these.

So this is project is really still in the ideas phase, but I'd welcome any thoughts or suggestions.

Questions are:

Do you know of any tested techniques for making recordings of this length, or have any other ideas on how it might be done?

Is there any equipment designed specifically for this?

Has this been done before?

or anything else you think is interesting.

Thanks,

Mark

ps. I think they record the Longplayer installation continuously, so that's on my list to investigate already.

Edit 08/07/12

Thanks for the responses and interest. I think it's clear that I need to do some more research around this, and as it's such a commitment in terms of time it's worth getting right. I'm also wondering whether if done correctly this could also have scientific value, so I'm going to try and investigate that further. Similarly, options for mics etc need to be talked through with the beekeepers to find out what's acceptable for the bees.

The recording device is a separate and interesting issue though, as there seems to be no dedicated low cost option. Matt Glenn's suggestion of using a Sound Devices recorder is probably the most solid idea, but I don't think this project will have the funds for that solution. But forgetting my particular problem with the bees for a moment, I've been wondering what the reason is for? Are people not making long recordings because there's no call for them, or are they not doing it because the tool required is not easily to hand? Would the device inspire the use?

I like the idea of a device being designed to do this which you can build yourself from off the shelf parts that is relatively cheap and adaptable. I think it could potentially make a good open source project. I've also been looking for a reason to get involved with the Raspberry Pi for a while now and this could be it.

So breaking this down into components of a system, at the core would be a flexible recording system controlled by editable/configurable open source software. The spec could be something like this:

• Raspberry Pi running a minimal Linux distro on SD
• Custom PD patch for control and recording
• USB hub
• 2tb HDD (Raid1 over USB?)
• Basic USB audio interface
• Set of preamps for microphones
• LAN or USB 3g modem for streaming and/or remote desktop/control
• Power supply
• 4-5" touchscreen for display/input

Now, I'm not 100% sure about this, but with careful choices I think you could power that from a battery and you could charge the battery with a solar panel. Build this into a case and you have a system which can be installed and left to run independently. Build a weatherproof case and you can leave it anywhere (although I'm not sure about protecting the mics from rain.. any ideas?). By adapting the casing and mic setup you could then create different designs for different purposes, for example you could add a parabolic dish for focused recording of one point or add two hydrophones and record a whole season in a pond or even build it all inside a binaural head...

If you're using PD for this there's also potential for expanding the system to take input and respond to external factors. Add an arduino talking to PD and you can use sensors to detect rain, temperature and sunlight, move parts of the system (like mic positioning) with servos or other motors for example. Even record other data as text. Think recording robot.

I can't say that I think I can do this myself – I've had some experience with Linux, PD and some basic Arduino interfacing, but my knowledge of electronics is pretty basic for one. But it could be a community project run from a Wiki which anyone could contribute to. Perhaps of interest to design students as well as sound recordists. The Wiki could suggest tested parts and configurations, provide install guides and software etc.

Well I'd love to hear your thoughts on this. Is it possible? useful? Do you see any serious problems? solutions? Would you contribute?

Answer 1

This is an awesome idea!

While I don't know of another project with a similar goal, I believe it was @Jay who turned me on to this live stream of underwater sounds of the Antarctic Ocean. It doesn't address your question directly, but it does make me think of a few ideas.

Perhaps instead of a lav or contact mic, you look into a hydrophone such as the H2a from Aquarian Audio? I've read much about their rugged usefulness on this and other boards and have heard of people freezing them in ice, burying them in the mud, and cramming them into cracks of trees while chopping them down. It might sound really awesome encased in wax! Plus, you don't have to worry about the bees, honey, or wax gumming up the works. Aquarian also has a custom shop if you find their stock mics aren't robust enough.

Another thing to consider if you do go the computer route might be a remote access app so you can log in remotely and check if a trip to the site is necessary between scheduled visits. This of course depends on where your hive is located. But if you have the net access, it may be beneficial.

Also, just as in the Antarctic project, a simple live stream would be nice. It would allow you to listen in over the course of the project to determine how your mic placement is working out, and you could post the stream so we can all listen to busy bees while we work. ;)

Answer 2

It might be possible to code something in Max/MSP or PureData that could handle the recording side for you on an always on computer. It would definitely take a bit of work, but you could set something to have the program run a cycle of [start and stop recording file on the hour]. It wouldn't be that hard to set up the file naming (also automatically handled by your coding) to be based on the date and time of the recordings either. Neither programs require a lot of overhead to run, so you could conceivably pull this off on a reasonably priced laptop or desktop.

Answer 3

The Sound Devices 700 series recorders allow you to record directly to a hard drive via firewire. I own one myself and I can tell you that they're reliable as clockwork. Get a big enough hard drive and you could feasibly keep up with the amount of data you need. I had to check to see if wav files are still limited to 2GB, but according to Sound Devices the maximum file size is actually 2TB with their formatting. At stereo 24b/48k, you'd only have to restart the recording 4 times to get a year's recording time... dang.

This is a really cool idea! I can't wait to hear what you capture. ~Matt

Answer 4

I guess another thing to think about is the editing or even using the file. Is there actually an editor that can open .wavs that are a year long? I think Pro Tools caps (or at least it used to) at 11 and a half hours or something similar. I'm not sure what other editor's limits are though.

There might even be a way to edit large files without actually opening them in a timeline, although I"m not familiar with it if there is.

Cheers

Answer 5

One thing is for sure - the formatting on SD disks is FAT32, meaning that the maximum filesize is going to be just under 4GB. SD does - however - do a good job of concatenating recordings across a number of files. I think the space constraint is more to do with the size of the volume than the size of the individual files. With a long, continuous recording, I would expect to find multiple files that will concatenate together on a timeline to produce the original continuous recording.

I think that another thing you should consider is that you should do some test recordings to determine spectrum first. You state that you're going to record at 48kHz/24bit but this may or may not cover the complete spectral range being emitted. For instance we generally record at 48kHz because of the range of human hearing however a lot of the sound sources we are dealing with emit sounds at higher frequencies than humans can hear. Determine whether you need to record at 48kHz or a higher sampling rate - then determine the impact on the storage space requirement.

I would also recommend separating the recording device from the microphones using a remote preamp and converter. This will have an impact on the powering source, but getting the recording device itself into a space where you can easily monitor and maintain it would be a consideration. There are many remote preamps and converters that will support connection either by fibre optic or UTP cable.

Answer 6

Since this project is so ambitious, I thought I'd share some ambitious ideas to go along with it.

I have only heard about them myself, and someone briefly mentioned them here, but I wonder if fiber optic mics would be useful in this scenario. They're essentially immune to noise and interference, robust, resistant to environmental changes in heat and moisture, and can go a realllly long distance (like, kilometers). You could set them in the bee hive and have your recorder back at the house (assuming there's a building around). That would take care of the need for a complicated and potentially sketchy weatherproof enclosure, solar powered this and that.

http://www.optoacoustics.com/industrial/optimic-microphones#general

This way you could also even have your recording computer connected easily to the internet, and set up an automated script to offload daily recordings to an FTP server, for remote access as well as to conserve local drive space, so you wouldn't need a massive disk on site.

I also second the idea of the audio time lapse as a means of presentation. I think it could make an interesting piece of visual art, however ironic that may be.

Answer 7

It's an interesting question that I think can be better addressed if you define your goals a little in a little more detail.

For example - what's the end use of the recordings going to be?

If it's primarily to generate a spectrogram I'd suggest going mono instead of stereo to save on disk space.

If it's for listening, then in what context? As a long form installation art piece? Even then you'll have to do some dramatic editing and/or compression I think. I like the idea of a livestream as a way to distribute for listening purposes though.

If it's for manipulation, then is there a reason to record the months of inactivity during the winter season? Why not start near the end of winter and end once they're all back in hibernation?

For what it's worth, I've developed a method of sonic timelapse that uses scripts and a combination of edits, time compression and overlaps and gets pretty high quality results (far better than time compression or editing alone). I'd imagine such a thing could be adapted to your project so that the entire year could be experienced in a matter of hours.

Actually, this project is begging to be timelapsed now that I think of it. If you had a webcam or GoPro clicking say 3-8 pix per day you could probably blend that with a sped up recording to truly get the effect of change that you seem to be looking for.

Answer 8

Interesting question: you may get more answers though if you ask it on the Nature Recordists board.......

Answer 9

This is an AWESOME idea - both the long recording of just bees, and even more so the construction of a device to make long recordings of anything and everything else.

I don't think I have much to add in terms of hardware/software (PD would definitely be the way I'd go with this - not sure about the hardware, though), but I think a really interesting application for such a 'long recording device' would be for acoustic ecology/soundscape studies. I liked the suggestion somewhere above about collecting the 'cycles of the planet' over a year, but what I think would be really interesting would be to monitor the changes in those yearly cycles over an even longer period, and so, map out broader changes in the soundscape of a particular location. (eg. imagine how interesting it would be compare the variations in the soundscape of, say, central London over the course of the last 25 or even 50 years). It would be even more interesting to also be able to compare the variations of different locations around the planet.

Of course, this sort of thing would be ambitious: you'd want a device that records pretty much indefinitely and requires minimal maintenance (the solar-power idea is great), and a community of operators. Though I wonder whether it might be of interest to Universities, libraries, etc. Or it could form some kind of broader community-led project into the planet's changing soundscapes...

I'd be on for developing such a device, anyway! Sounds fun!

Answer 10

Do you know of any tested techniques for making recordings of this length, or have any other ideas on how it might be done?

Yes. We've successfully recorded weeks of video plus audio on a Raspberry Pi using a WD PiDrive hard disk for storage. I've tested multiple microphones for use with the raspberry pi and only two of them would record at an acceptable level. One was built in to a headset and not appropriate - the other was this contact mic/guitar pickup costing about $2: (search eBay for "Piezo Contact Microphone Pickup for Guitar"). The sound card (search eBay for "White USB Virtual 7.1 Sound Card"), costs a dollar. Attach the contact mic inside the frame. The bees will probably cover it in propolis but it will record vibrations through the wood like the soundbox of a guitar, so that shouldn't matter.

Has this been done before? Yes - http://annemariemaes.net/publications/the-sound-beehive-experiment/ (search for "The Sound Beehive Experiment") - very similar to what we're doing ourselves but haven't documented online yet.

Answer 11

Not so much an answer as a thought, but bees produce a lot of propolis in the hive, so you might find any equipment left within gets slowly sealed over, which will alter the quality of your recordings. It might also slowly insulate your other equipment in there, which might cause it to overheat. I've thought about this before after a couple of my colleagues attended a seminar on sensors and 'The Internet of Things'. We have a beehive at work, and inspections are carried out in the traditional manner by lifting the top off and carefully prizing the frames apart. During the summer some frames are removed and the honey is extracted, but during the Spring months inspections are carried out simply to check the state of the colony. I was wondering if more of this could be carried out remotely, especially in cold weather, as the bees then have to work harder to get the hive back to an optimal temperature.

Moreover by monitoring the audio within the hive, generally the pitch and volume of the bees' humming, alongside hive temperature and number of movements in and out of the hole at the front of the hive (I thought a white LED shining onto an LDR hooked up to a counter of some sort might work), you might collect enough data over time to be able to give an early warning prior to a hive collapse, and the apiarist could move in and save the colony. For example a hive that is at risk of collapse may become steadily quieter and cooler in the 24 hours prior to a colony collapse.

Unfortunately I only know the rudiments of sound recording, bee keeping and remote sensors, so I've not taken this idea anywhere.