I have a few microphones and combined with different recorders, don’t sound consistent. Issues being all the elements that make up the chain between the noise itself and the recorded sound. That’s going to include the noise source, background noise (that the brain often filters out), the microphone, the microphone cable, the recorder and any compression artefacts added to the recording if you are not using an uncompressed file format.

I have also been trying to unravel the technical details to understand why. So I used a little bit of Google and quite a bit of reading. In the end I didn’t really get to grips with it, so I will carry on knowing how my gear works in which situations (which is probably the best approach). Here are a couple of snippets of interesting details/science/etc.

I found this as a good start point, a rule of thumb for microphones.

Max SPL = The point where the mic distorts, or clips the waveform. [More=better]
Self noise = The amount of noise the mic creates all on it’s own. (i.e. hiss). [Less=better]
Dynamic range = The range between self noise and Max SPL. [More=better]
Signal to Noise ratio = The range between self noise and a reference signal. [More=better]

Reference

• http://homerecording.com/bbs/showthread.php?t=276820

So a good start.

But then you don’t always get this data from all microphone specifications. So this explains on how to fill in the gaps.

Some mics may only provide a specification for signal to noise ratio instead of self-noise. Higher numbers are better on this one. Fortunately you can figure out the self-noise from this if they provide enough info. Typically microphone signal to noise ratio is related to a reference acoustic level of 1 Pascal (Pascal is a unit of pressure and relates to “Sound Pressure” [SPL]). A typical rating might be 70 dB, 1 Pa @ 1 kHz. 1 Pascal is equivalent to 94 dB SPL. So, for a S/N ratio of 70 dB you can assume the self-noise to be 24 dB. There are a variety of testing methods, standard, and units of measurement that can come in to play to further muddy the waters, but you will find that most reputable microphone manufacturers will generally give you enough information in a standard enough format to make a comparison.

Reference

• http://www.sweetwater.com/expert-center/techtips/d–01/06/2004

So what does this mean then? For nature recording, getting close to the subject with any thing will give a good result. Further away from the subject will require more recording gain, therefore potentially more recording hiss.

What I usually do is record at relatively low levels quieter sounds and then use an audio file editor (usually Audacity) to amplify. when I can hear hiss being introduced I usually then filter out with a low pass filter. If the recording is used on its own you can hear the difference, but if mixed in with something else, rarely heard.

For mechanical sounds, things are a bit easier. High pass and low pass filters can be used to cut our unwanted sound and also enhance wanted sound. Using a spectrum analyser is sometimes useful for finding out these frequencies. Additionally to high/low pass filtering, EQ can be used to address specific noises on specific frequency bands.

The language took a hour or 2 to get used to and came up with the following.

set file_ to choose file
set file_alias to (file_ as alias)
tell application "Finder"
	get properties of file_alias

	set d to (get modification date of file_alias)
	set t to (time string of d)
	set h to (hours of d as string)
	set y to (year of d as string)
	set m to (month of d as integer as string)
	set dy to (day of d as integer as string)
	set ss to (seconds of d as integer as string)
	set mm to (minutes of d as integer as string)

	-- fix padding of day
	repeat while length of dy < 2
		set dy to ("0" & dy)
	end repeat

	-- fix padding of month
	repeat while length of m < 2
		set m to ("0" & m)
	end repeat

	repeat while length of h < 2
		set h to ("0" & h)
	end repeat

	repeat while length of mm < 2
		set mm to ("0" & mm)
	end repeat

	repeat while length of ss < 2
		set ss to ("0" & ss)
	end repeat

	set f to (get folder of file_alias as string)

	set long_string to y & "-" & m & "-" & dy & "T" & h & "h" & mm & "m" & ss & "s" & ".wav"

	display dialog (long_string)

	set name of file file_alias to long_string

end tell

Current limitation is that it can only be run once, on a per file basis, but this gets me to renaming files like STE-012.wav to 2009-04-10T14h07m24s.wav, so a good start point.

Update

The following allows a batch process to be done using a multiple file selection. From here it would be good to use a temporary drop folder with automation attached. Maybe another day.

-- open a dialog with multiple selections enabled
set file_ to choose file with multiple selections allowed

-- set a boolean to skip verify action
set global_OK to false

repeat with this_file in file_

	set file_alias to (this_file as alias)
	tell application "Finder"
		get properties of file_alias

		set file_type to (get kind of file_alias)

		if (file_type is "WAVE Audio File") then
			--display dialog (file_type)

			set d to (get modification date of file_alias)
			set t to (time string of d)
			set h to (hours of d as string)
			set y to (year of d as string)
			set m to (month of d as integer as string)
			set dy to (day of d as integer as string)
			set ss to (seconds of d as integer as string)
			set mm to (minutes of d as integer as string)

			-- fix padding of day
			repeat while length of dy < 2
				set dy to ("0" & dy)
			end repeat

			-- fix padding of month
			repeat while length of m < 2
				set m to ("0" & m)
			end repeat

			repeat while length of h < 2
				set h to ("0" & h)
			end repeat

			repeat while length of mm < 2
				set mm to ("0" & mm)
			end repeat

			repeat while length of ss < 2
				set ss to ("0" & ss)
			end repeat

			set f to (get folder of file_alias as string)

			set long_string to y & "-" & m & "-" & dy & "T" & h & "h" & mm & "m" & ss & "s" & ".wav"

			if (global_OK is false) then

				set question to display dialog "Renaming " & file_alias & " to " & long_string buttons {"OK", "OK for All", "Cancel batch"} default button 1
				set answer to button returned of question

				if (answer is "Cancel batch") then
					display dialog "Cancelled" buttons {"Understood"}
					return
				end if

				if (answer is "OK for All") then
					set global_OK to true
				end if

			end if
			set name of file file_alias to long_string
		else
			display dialog ("Not a wav file, skipping")
		end if
	end tell

end repeat

I wonder if this can be used to any advantage when recording or processing found sound?