Since you use an area unit to describe the set-up, I'm assuming this is a 2-dimensional type of array, but you give no idea of the number of microphone elements. Similarly, use of the word "array" implies that you will be performing some type of corellation of the outputs of the various transducer elements, so simultaneous multi-channel recording and/or processing will be required. An additional complication will be the need for weather-proofing and how this might affect the captured sounds.

I'm also assuming you are talking about using omni-directional microphones, and that the distance between the microphones is such that any sound of interest will be picked up by more than one microphone.

Is your prototype for demonstration of the acoustic principles, or to show that you can put together a microphone array that has some chance of standing up to the elements?

Give us some more idea (for both prototype and final unit) of the size of the scheme, the number of microphones, the sensitivity needed, the recording method and any other constraints.

Interesting project. Good luck!

Very interesting.

Are you trying to record for the sake of audio or trying to gauge noise-levels over a matrix? In other words, do you care what it sounds like, or are you just mapping decibel levels?

And if you could please give us your definition of 'cheap'.

Ok, so to give you some more information...We are looking to understand how birds respond to noise in their environments. The idea is to create 3D sound maps of noise with bird territories overlaid on those maps. With these maps we can answer many biologically and evolutionarily important questions.

For our prototype setup, we are thinking of just mapping sound over a single territory (1-2 hectares) and then moving the setup to another territory. I am thinking that between 8 and 16 microphones would be enough. They would all have to be synchronized and recording simultaneously to an interface that can store the data (National instruments for example).

The more I read, the more I figure out how this is going to work, but some of the finer details still elude me. We are interested in noise between 1000 hz and 10,000hz if that helps. I could use the most help in choosing omnidirectional microphones that are relatively inexpensive (less than $100 if possible) yet are accurate and sound enough to provide quality results.

Thanks for the input so far.

This is going to be a tough assignment! Even if you take the smallest area you mention (1 hectare) and the most microphones (16), that still leaves a minimum distance of over 33m between mics if they are arranged in a rectangular grid. I don't think many of the live sound folks on this forum would consider that 33m between mics gave adequate coverage, given standard sensitivities, but it may just work for you in this project.

In terms of the type of microphone that could be used in an outdoors for this type of work, I suggest that you look at Pressure Zone types (PZMs), originally patented by Crown, but now produced by a number of companies since the patent expired. These have the advantage that they can be placed directly on the ground, so would not need stands. They have a useful fall-off of their low frequency response if you do not extend their plate area, and, importantly, they also have a chance of being more weatherproof than conventional microphone types.

I suggest that you consider purchasing a couple of PZM mics and perform experiments with them spaced at 30+ metres. You want the PZM type that can be powered from the recording device rather than having to rely on internal batteries, so, for example, the Crown PZM185 might be a place to start. The list price of this one is around $200, but they are likely to be available cheaper than that, especially in quantity.

You could have great fun laying out the mics between two trees 50m apart and stringing a line overhead, towing a small MP3 player on a pulley along the line and recording the microphone responses.

If you don't already have an audio interface for recording the microphone outputs, then a battery device such as the Zoom H4N could be used to capture two external channels, and can be later plugged into a computer to transfer the recordings. The H4N has the disadvantage that it's not the lowest noise unit available, but if you find you can get reasonable results with it, then the recordings would be even better using a lower-noise multi-channel interface. An additional factor to take into account is that the H4N would not easily form a part of the project once you go to more than two mics, so you would have to regard it as part of the cost of a feasibility trial.

In the $100 range, I think it's hard to beat the Electro-Voice 635 for an almost indestructible omni-directional mic.

This [[url=http://[/URL]="http://kenai.fws.go…"]Kenai project[/]="http://kenai.fws.go…"]Kenai project[/], used decibel meters in conjunction with audio microphones since noise-level was their main focus.

SOME IDEAS AND THOUGHTS

Unregistered, post: 360881 wrote: Hi,

I am going to start off by saying I am a phd student in evolutionary biology,....

[SO THEN JUST WHY ARE YOU DOING ACOUSTICS ?]

I am trying to design an outdoor microphone array over several acres to simultaneously record sound levels of urban environments.

[WHAT PRECISION RESOLUTION DO YOU NEED? SPACE AND TIME DIMENSIONS -- FREQUENCY AND PRESSURE TOO ]
[WHAT ACCURACY DO YOU NEED]

....My collaborator and I are trying to build a prototype system (on the cheap) ...

then apply for a NSF to build a better system for next year.
[I SEE ANOTHER GOLDEN FLEECE AWARD FOR WASTING *MY* TAX DOLLARS]

The problem is, I don't know where to start looking for the components online or what I need to build this.
[UNTIL YOU DEFINE WHAT YOUR OBJECTIVE IS AND HAVE A DESIGN TO DO IT YOU WILL NEVER GET AN ANSWER TO THIS]
[GOOGLE IS YOUR FRIEND]

Any input would be greatly appreciated. ...

DEFINE PRECISELY WHAT YOUR OBJECTIVE IS

DETERMINE THE PARAMETERS YOU NEED EG RESOLUTION IN TIME SPACE LOUDNESS ETC.

START SMALL TO PROVE FEASIBILITY

LOOK AT ***ALL*** ALTERNATIVE SOLUTIONS BEFORE PICKING ONE

Do these all need to be synchronised ?
Mono or stereo ?
Resolution = inches feet yards miles parsecs?
WHERE ARE YOU GOING TO PUT ALL THESE THINGS ?
What resolution for sound pressure level ? 0.1 db ? 1 db 10??
Time resolution ? nanoseconds, seconds, minutes, hours, years, ...?
What freq resolution ? Bandwidth = 1 cps,10, 1000cps , at what center freq ? Earthquake rumble? human audio? bat radar ??

How you automate and analyse data capture will be a much bigger problem than merely using some mikes to determine sound level.
But those usually require special mikes that are calibrated for SPL use.

WHEN YOU CAN ANSWER THE BASIC QUESTIONS THEN WE CAN POINT YOU TO SOME EQUIPMENT THAT MIGHT MEET YOUR NEEDS

Basically what you are describing is a different use of sonar. And I'm not talking about a software of any sort. You're trying to utilize sonic ranging principles. This is already in use in the Washington DC downtown area. However it's being done with more than just a couple of microphones. They're not listening for birds however. They're listening for gunshots. When a gunshot is detected, the time differentials between these multitudes of microphones can be used to establish the point of origin. So no actual synchronization is utilize its the lack of the synchronization that makes it work. So I totally believe that your research concept is valid. The problem however is in knowing what the best way would be to spread your microphones out over a distance and then having the microphones perfectly synchronized. Well that's easy also. Knowing that any kind of electricity travels through our wires at 186,000 mi./s all you have to do is make sure that all of your microphone lines are coming back to any particular recorder, all of the same length. Now everybody might be laughing. Go ahead and laugh, I did. That is until in 1981, when I went to work for NBC-TV in Washington DC did I discover that our genius Chief Engineer the now late Michael Galvin was entrusted to rewire the entire station for color television. There was a big problem with color television in that if the timing and synchronization wasn't exactly right, you couldn't get all the colors from camera to camera to match. So when all of the hundreds of miles of wires were installed in the building, there was a lot of wires that were longer than they needed to be because every cable sharing every video signal, no matter how long, were all cut to within 1/4"! No joke. This works. And as a result, color timing of NBC TV Washington DC was always spot on. And this is why it was so exciting for me, even though I had been well mentored, to be hanging out with all of these guys that were there, before the beginning. Is there a before the beginning? I think there has to be? I think is what the physicists have always been talking about. The before the before.

And, to make this all work for you, you only have to understand what the special content is of what you're trying to time. That way your concern for high fidelity versus low fidelity sound is not a factor. This is not about making a recording. This is about measuring the speed of light at different points in time. It should be easy for anybody with a PhD to do. Oh, that's right, you don't yet have your PhD. Sorry.

I have a PhD in going to the bathroom
Mx. Remy Ann David

I haven't forgotten about you guys. My old laptop died and I am just getting up and running. I'll reply to posts within a day or two. Thanks for all of the input so far!

Not sure why the original post was closed, but I have some updates. I conducted a test run of this idea with [="http://www.audio-technica.com/"]Audio-Technica[/]="http://www.audio-te…"]Audio-Technica[/] omnidirectional microphones attached to laptops. I placed them on the corners of a 20 x 20m square. If you would like to see all the details of the setup, including weather conditions and exact measurements, email me at zjrj5@yahoo.com. I can send you a copy if that helps you answer my questions.

Anyways, I am looking at the recordings in Adobe Audition 3.0. I am trying to time synchronize them. After I set all the microphones to record, I produced a beep at a known [URL="http://www.spl.info within the array to signal the start of official recording. I also did this at the end of recording. When I look at the .wav file for computers 1 and 2 to see how long the recording ran for, I am off by .044 seconds, which translates to 13.6m of error with the speed of sound being 331.6m/s yesterday (taking into account, temperature and humidity). This amount of error is too large.

My question to you is two part. Is there a way to make these 4 .wav files line up (time synchronized) exactly? For our formal research, we won't have a perfect square that we can produce the "beep" right in the middle. We will have a polygon of sorts with microphones scattered throughout. Maybe when I am clicking in Audition, I am not clicking on the exact start and stop points of the beeps. Suggestions? For our formal research, we plan on using the Zoom H2 Handy Recorders instead of the Audio-Technica and laptops. This system seems to be the best for our money and what we need it for. Does this change anything as far as synchronization ability? If and when we get more funding once we collect our preliminary data this spring/summer, we would like to use the National Instruments interface to help with this issue.

Any and all help would be greatly appreciated. Thank you ahead of time!

unregistered wrote: I conducted a test run of this idea with Audio-Technica omnidirectional microphones attached to laptops. I placed them on the corners of a 20 x 20m square. If you would like to see all the details of the setup, including weather conditions and exact measurements, email me at zjrj5@yahoo.com. I can send you a copy if that helps you answer my questions.

Anyways, I am looking at the recordings in Adobe Audition 3.0. I am trying to time synchronize them. After I set all the microphones to record, I produced a beep at a known SPL within the array to signal the start of official recording. I also did this at the end of recording. When I look at the .wav file for computers 1 and 2 to look for how long the recording ran for, I am off by .044 seconds, which translate to 13.6m of error with the speed of sound being 331.6m/s yesterday (taking into account, temperature and humidity). This amount of error is too large.

My question to you is two part. Is there a way to make these 4 .wav files line up (time synchronized) exactly? For our formal research, we won't have a perfect square that we can produce the "beep" right in the middle. We will have a polygon of sorts with microphones scattered throughout. Maybe when I am clicking in Audition, I am not clicking on the exact start and stop points of the beeps. Suggestions? For our formal research, we plan on using the Zoom H2 Handy Recorders instead of the Audio-technicas and laptops. This system seems to be the best for our money and what we need it for. Does this change anything as far as synchronization ability? If an when we get more funding once we collect our preliminary data this spring/summer we would like to use the National Instruments interface to help with this issue.

Any and all help would be greatly appreciated. Thank you ahead of time!

Are you saying that you used 4 unsynchronized laptops to record the 4 mics in the 20m array? These laptops are not going to sample at exactly the same rate, so you will not get phase-accurate recordings. To do that, you must either have a single multi-channel interface to which you connect all the microphones, or you have a recording interface per microphone-computer combo, each interface having a clock source cable back to a common sampling clock generator unit. Any combination inbetween these two would of course give the same result, e.g. a two-channel interface with computer and clock cable per two microphones.

Unless you use this or an equivalent technique, you cannot get any sort of phase coherency between the microphones and hence begin to calculate positional information using interchannel delays and the velocity of sound. Synchronising the start times is not enough, and even giving start and finish gunshots to line up in your recordings does not compensate for different clock drifts in the individual recorders due to varying cloud cover and similar effects.

This does not bode well for recorders like the Zoom that lack external clock inputs. Consider this point carefully before you splash out large amounts of cash on mics/recorders for each capture position in your array.

That is exactly what I am saying. Thank you very much for the input. This is a major issue now. I am going to have to redesign this system clearly. I looked into the National Instruments system, but it is costly. My start up costs can't be too high. Any ideas for a 8 omnidirectional microphone set up that meets your requirements for phase-accurate readings and my requirements for as low cost as possible, yet functional?

How are you assigning costs between the trial (4/8 channels) and the final system? Do you have to build a working trial and then use the components but add to them to get to the final system size? Does it all need to be battery-powered?

For a phase-coherent trial (prototype) system up to 8 channels, I would go for a multi-channel pre-amp with FireWire interface. Position the interface with the laptop computer roughly central in the array (if possible), and run cables off to the microphones.

Have you chosen the type of microphone yet? For the money, I think you should at least look at something like the Radio Shack Omnidirectional Boundary Microphone (PZM) for this type of project.

Boswell, post: 364631 wrote: Are you saying that you used 4 unsynchronized laptops to record the 4 mics in the 20m array? These laptops are not going to sample at exactly the same rate, so you will not get phase-accurate recordings. To do that, you must either have a single multi-channel interface to which you connect all the microphones, or you have a recording interface per microphone-computer combo, each interface having a clock source cable back to a common sampling clock generator unit. Any combination inbetween these two would of course give the same result, e.g. a two-channel interface with computer and clock cable per two microphones.

Unless you use this or an equivalent technique, you cannot get any sort of phase coherency between the microphones and hence begin to calculate positional information using interchannel delays and the velocity of sound. Synchronising the start times is not enough, and even giving start and finish gunshots to line up in your recordings does not compensate for different clock drifts in the individual recorders due to varying cloud cover and similar effects.

This does not bode well for recorders like the Zoom that lack external clock inputs. Consider this point carefully before you splash out large amounts of cash on mics/recorders for each capture position in your array.

How much with clock drift really play a role? I think if I get a good enough impulse noise to signal the start and end of recording, I might be able to get it to work. I just have to bundle the H2's together, play the impulse noise and then spread them through the area. How does that sound to you?

The clocks will not be easy to sync at all since they will not share any audio spikes to assist alignment.

segerfan83, post: 364682 wrote: How much with clock drift really play a role? I think if I get a good enough impulse noise to signal the start and end of recording, I might be able to get it to work. I just have to bundle the H2's together, play the impulse noise and then spread them through the area. How does that sound to you?

Not good.

If the clock drift between any two recorder units were 100ppm, then after an hour there would be 360ms discrepancy, corresponding to over 120m of sonic distance.

Even if you were to gather up the recorders at the end of the recording before hitting the stop button and record another beep, you still have the problems of (a) adjusting the recorded waves by the non-integer sample amounts that would be needed to correct individual events within the recording, (b) guessing at non-linear relative clock drift due to temperature and other factors.

I've had many academic papers to referee in my time, and several of them have been brilliant at their specialised topic but have fallen at the first hurdle of technical consistency. What I'm saying is that there are easy ways of getting this aspect of your project correct from a technical point of view, and, although they may not appear to offer the lowest-cost solution, battling at the result analysis stage with errors due to poor experimental implementation is not the way to conduct good science.

I don't want to dissuade you from going ahead with this interesting project, but I would be happier if I knew you recognised that some apparent cost savings have, in reality, the effect of compromising the whole underlying technical integrity of the method.

Is everyone sharing these same sentiments? I need to know that this isn't a concern of just one person, but its an actual 100% issue with this sort of setup. We are thinking of switching to this setup. [[url=http://[/URL]="http://www.wildlife…"]Song Meter SM2 for recording birds[/]="http://www.wildlife…"]Song Meter SM2 for recording birds[/] but the costs are more and we need to be 100% sure that switching to something like this is the difference between sound science and bad science and not the difference good science and slightly less good science to justify the costs. This isn't a knock on the opinions already, just making sure I cover all my bases.

You would do well to heed Boswell's advice - they are not sentiments but valid engineering. Not having precisely correlated results will cause you severe and intractable headaches in processing the data. I have done work in marine Geophysical Exploration where there can be arrays of several hundred hydrophones (marine microphones). Yours is a very similar application technically. To get accurate results it is absolutely essential that the data collected is 100% correlated or the results would be effectively useless.

The 2-channel SM2 Song Meter is a nice unit for what it is designed to do, and it may well have the edge over domestic products when it comes to environmental factors, but you would have to make a good technical case for it not having the same synchronising problems as Zoom H2s.

You can't get away from the fundamental requirement of synchronicity for experimental work of this type. This requires you either to log all the data on a single multi-channel recording system, or to distribute a centrally-generated synchronising clock to individual distributed recorders, which must be of a type that can follow a clock input.

I'm not trying to be difficult, but I am trying to save you from making an expensive mistake.

Boswell, I don't think you are being difficult at all. I am appreciative of this information more than you realize. You have to understand where I am coming from. I am a evolutionary biologist with no background in this area. That's why I am being so cautious with my next steps.

As far as the new setup I presented, these devices are synchronized with a common GPS clock that the company's software uses to provide this synchronization. Doesn't that solve this issue?

As far as clock drift, what are some of the causes of this? I am assuming temperature, but how much will this affect digital data capture rather than old fashioned tape data capture?

In the SM2 manual I looked at in some detail before posting last time, it briefly mentioned GPS synchronisation as a future option but gave no further details. After reading your last post, I dug around a bit more and found this, which indicates that the option is indeed now available, and states that it gives synchronicity to within 1 millisec (i.e. about 1 ft sonic distance).

Use of this GPS option to a large extent removes the fundamental reservation I had about your unsynchronised proposal, as it acts rather like a reduced-accuracy version of the common clock distribution method I mentioned. If the 16-bit conversion and other specifications of the SM2 unit are acceptable to you, it does look as though this company (Wildlife Acoustics) has experience in technical aspects of the area of science you are proposing to move into, and I wish you all the best with the project. Let us know how it all works out.

The Why is of little interest for you in regards of performing the task.

What has been said. Get a multi input mic preamp for 8 or 2 for 16 mics that has digital out connection and plug into a notebook DAW or MT hardware recorder.
Hook up the mics with a cable and press "record". No sync problems, no drift, ....no pain.
If audio quality is not the highest priority, but budget, have a look at something like this [[url=http://[/URL]="http://www.thomann…"]BEHRINGER ADA8000 PRO-8 DIGITAL - Thomann UK Cyberstore[/]="http://www.thomann…"]BEHRINGER ADA8000 PRO-8 DIGITAL - Thomann UK Cyberstore[/]
Boswell already mentioned the use PZM mics. Those should be your choice. Apart from that and again, if budget is ultra tight, .. you get something recorded at 1 kHz to 18 kHz with almost any cheap mic.

Now, shoo-shoo,...off in the forrest and molest the little birdies..
;-)

Record all the tracks simultaneously into one session of Audition. Recording onto different computers introduces any number of variables in latency, etc. You're going to need some long cables, and an interface capable of receiving all the necessary inputs and sending them to Audition. Many firewire mixers would be able to do this. In order to set different tracks of Audition to record simultaneously, you need to make sure that each track has a unique send from a channel of the mixer. Then enable all the tracks you're using to record. Each track (as long as a discreet track has been assigned to every mic) should represent the recording from one microphone's perspective. Unless you are sure that the sound source for your initial beep is equidistant from all microphones, there is going to be a margin of error due to the difference in distance for each microphone versus the speed of sound (relatively constant for any given day). If you don't care about a "natural" representation, but just want synchronization, just take the "beep" you play after you start recording the clips and line each mic's beep up in Audition. Using the hybrid tool (press V to select it), right-click on a clip and drag it for coarse timeline shifting. For more fine movements, select a clip and use Alt+left/right arrows to move back and forth ever so slightly. Zooming in to a ridiculous amount and lining up the beeps from each clip should give you the results you are looking for... even if it isn't a "natural" representation of the sound space. Hope this helps.