Microphones and Quality of Recording.

[Alyrium Denryle]

So, I am trying to set up a system to remotely record frog choruses in ponds. I am restrained to about 100 USD per unit, of which there will be 18. Units will consist of a digital recorder, omnidirectional dynamic microphone (this one designed for this purpose), and a solar charger, placed in a weatherproof casing.

My question is this. Say I have an audio recorder with a frequency response of 70 hz to 20 khz, capable of recording in .WAV or .MP3 formats. If I use a microphone with a higher frequency response (say 20-20khz), will the recorder limit the capabilities of the microphone? Would it depend on how the recorder encodes its data perhaps?

Note: Assume I have shut off all noise cancellation features.

[Dragon Angel]

Does that recorder have its own internal microphone? I believe that number would refer to something like one of those.

What kind of audio connectors will you be using: TRS or XLR? It should be able to faithfully record whatever inputs your microphone will catch, but XLR cables are generally considered better for higher fidelity and less noise.

[phongn]

My question is this. Say I have an audio recorder with a frequency response of 70 hz to 20 khz, capable of recording in .WAV or .MP3 formats. If I use a microphone with a higher frequency response (say 20-20khz), will the recorder limit the capabilities of the microphone? Would it depend on how the recorder encodes its data perhaps?

You'd have to get the actual response graph of the recorder or do a test.

[Alyrium Denryle]

Other quick question:

Anyone have any comparison (qualitative will work) between that data loss of using WMA or MP3 compression?

I ask because I misstyped in my original post. The recorders I have available can do MP3 or WMA, as opposed to MP3 and WAV, and I know both are lossy, but use different compression methods, but I have no idea which one loses more data in so doing. I need to feed these into a trained recognizer for frog calls, and would like to minimize data loss as much as possible.

Any help is greatly appreciated.

P.S. Thank you for the microphone help. I ran some tests and resolved that issue.

[evilsoup]

Well WMA is a a newer format, designed to overcome some deficiencies in MP3, which is quite old and really only used because of greater compatibility and inertia. I'd say go with WMA. Of course, since you are concerned about quality, make sure you set the recorders to their highest quality settings (highest bitrate).

[Dominus Atheos]

Other quick question:

Anyone have any comparison (qualitative will work) between that data loss of using WMA or MP3 compression?

Once again, do a test.

The problem with a comparison is that it all depends on exactly hardware codec chip is inside your audio recorder. Here is a listening test from 2005 that shows that the human ear cannot detect any loss in MP3, WMA, AAC, or Vorbis even at a 128Kbps bit-rate.

Your hardware recorder will not get those kinds of results, and the only way to tell which codec is acceptable at what bitrates is to do a manual test yourself.

[darkjedi521]

What's your microphones frequency response graph, and sensitivity over distance? That will have as much or larger impact on your recording that the recorder itself.

[Melchior]

the human ear

He's going to record frogs. A significant part of the recording will probably fall outside the audible range for humans and still have scientific merit - I would avoid compression standards that are going to discard everything that humans are not likely to hear for smaller size.

[evilsoup]

Both MP3 and WMA throw away a lot of data that is outside the hearing range of the human ear. WMA gets better subjective quality per bitrate (and I think has greater dynamic range), but both of them are lossy in that regard.

[Alyrium Denryle]

the human ear

He's going to record frogs. A significant part of the recording will probably fall outside the audible range for humans and still have scientific merit - I would avoid compression standards that are going to discard everything that humans are not likely to hear for smaller size.

Well, it is being used for identification, the stuff that is outside the human hearing range is a bit less important. For me, the concern is whether or not I will be able to automate identification via an analysis program. I think I should be able to do it, provided I can turn off all noise reduction options, bring in as much data as I can, and then use the same encoder to make the data set uniform. Then I can go out before my official field season, record individual calls, and build my recognizer, test it to make sure it can ID frogs in huge choruses... yeah

If I were analyzing the calls of individuals I would be screwed, but I think I am OK for identification purposes.