r - Finding and counting audio drop-outs in an ecological recording -

i trying assess how many audio drop outs in given sound file of ecological soundscape.

format: wave samplingrate (hertz):   192000 channels (mono/stereo): stereo pcm (integer format):   true bit (8/16/24/32/64):    16  

my project had 2 element hydrophone. elements different brands/models, , trying determine element preformed better in our specific experiment. 1 analysis conduct measuring how each element had drop-outs, or loss of signal. these drop-outs not signal amplitude related, in other words, drop-outs not caused maxing out amplitude. element or associated electronics failed.

i've been trying in r, program familiar with. have limited experience matlab , regex, opening trying programs/languages. i'm biologist, please excuse ignorance.

in r i've been playing around package 'seewave', , while i've been able produce pretty spectrograms (which, fair, context i've used package). attempted use envelope , automatic temporal measurements function within seewave (timer). got interesting, opposite results.

    foo=readwave("documents/dasbr/dasbr2_20131119$032011.wav", from=53, to=60, units="seconds")     timer(foo, f=96000, threshold=6.5, msmooth=c(30,5), colval="blue") 

i've altered values of msmooth , threshold countless times, that's fine tinkering. function preforms measuring duration between amplitude peaks @ given threshold. need either a) find samples in signal without amplitude or b) measure duration between areas without amplitude. can work either of outputs. want reverse direction threshold measuring, make sense? therefore sample below threshold trigger measurement, rather sample above threshold.

i'm still playing seewave see how produce data need, i'm looking bit of guidance. perhaps there function in seewave accomplish i'm trying more efficiently. or, if there anyway output numerical data generated timer, use 'quantmod' package function 'findvalleys' list of data gaps.

so yeah, guidance i'm requesting, oh data crunching gods.

cheers.

this problem sounds reminiscent of power transfer problems seen in electrical engineering. 1 way solve problem take rms (the root of mean of square) of samples in signal on time, averaged on short durations (perhaps few seconds or shorter). durations see low rms dropouts are. it's analogous vu meters see on audio amplifiers - indicate power being transferred speakers amplifier.