make.peaks: Extract a continuous peak-frequency contour from a...
In landmarkacoustics/SoundPoints-R: an R version of the SoundPoints algorithm

Description Usage Arguments Value Author(s) See Also Examples

At each time step of a spectrogram, this function finds the maximum-energy frequency, then identifies the band around that peak that holds a predetermined total amount of energy

1	make.peaks(spg, passband, amplitude.cutoff)

`spg`	a spectrogram object
`passband`	a two-element vector that defines the frequency range that will be searched
`amplitude.cutoff`	a single numeric value that defines the total area of each peak

a data frame in which each row corresponds to a time point of the input spectrogram. It has the following columns:

`time`	the time that the row describes
`min.frequency`	the lowest frequency in the peak (inclusive)
`max.frequency`	the highest frequency in the peak (inclusive)
`amplitude.sum`	the total of all the amplitudes in the peak's range
`mean.frequency`	the weighted mean frequency of the peak
`mean.amplitude`	the mean amplitude of the peak
`max.amplitude`	the peak's greatest amplitude
`amplitude.range`	the difference between the peak's maximum amplitude and its minimum amplitude

Benjamin N. Taft ben.taft@landmarkacoustics.com

spectrogram

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (spg, passband, amplitude.cutoff) 
{
    freqs <- attr(spg, "frequency")
    g <- freqs > min(passband)
    below <- sum(!g)
    g <- g & freqs < max(passband)
    peak.values <- data.frame(t(apply(spg[, g], 1, collect.energy, 
        amplitude.cutoff)))
    names(peak.values) <- c("lo", "hi", "sum")
    peak.values[, 1:2] <- peak.values[, 1:2] + below
    peaks <- data.frame(time = attr(spg, "time"), min.frequency = freqs[peak.values$lo], 
        max.frequency = freqs[peak.values$hi], amplitude.sum = peak.values$sum)
    for (i in 1:nrow(peaks)) {
        ix <- peak.values$lo[i]:peak.values$hi[i]
        amps <- spg[i, ix]
        peaks$mean.frequency[i] <- weighted.mean(freqs[ix], amps)
        peaks$mean.amplitude[i] <- mean(amps)
        tmp <- range(amps)
        peaks$max.amplitude[i] <- tmp[2]
        peaks$amplitude.range[i] <- diff(tmp)
    }
    return(invisible(peaks))
  }