fast.soundpoints: a fast and dumbed-down version of soundpoints

Description Usage Arguments Value Author(s) See Also Examples

View source: R/fast.soundpoints.R

Description

this version is faster because it only returns the landmark data frame, not the peaks or the spectrogram. it also makes some stupid assumptions about the sample rate of the sound, don't use it if it's not 44.1 kHz.

Usage

1
2
3
4
5
6
7
8
fast.soundpoints(file.name,
 passband,
 amplitude.cutoff,
 loudness.column,
 half.life,
 var.columns,
 var.weights,
 n.landmarks)

Arguments

file.name

the path to the sound file

passband

a two-element vector specifying the frequency passband

amplitude.cutoff

the total amplitude under a peak that determines its frequency range

loudness.column

the column of the peak matrix that corresponds to the amplitude you want to use

half.life

the decay rate of the weighted amplitude average used to discriminate between signal and noise

var.columns

the columns from the peak matrix that will be used to calculate landmarks

var.weights

the weights from var.columns that determine event detection. basically, you should write it as 1.0 / (critical differences).

n.landmarks

the number of landmarks in each set

Value

file.name

the name, without path, of the source file

landmarks

a data frame of landmarks, with one row per event

Author(s)

Benjamin N. Taft [email protected]

See Also

soundpoints

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
##---- 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 (file.name, passband, amplitude.cutoff, loudness.column, 
    half.life, var.columns, var.weights, n.landmarks) 
{
    spg <- spectrogram(file.name, 107/44100, 100)
    peaks <- make.peaks(spg, passband, amplitude.cutoff)
    initial.amplitude <- quantile(peaks$mean.amplitude, 0.85)
    amplitude.threshold <- log(10)
    time.res <- get.time.res(spg)
    rm(spg)
    peaks$smooth.amplitude <- smooth.average(peaks[, loudness.column], 
        time.res, half.life, initial.amplitude)
    not.noise <- peaks[, loudness.column] > peaks$smooth.amplitude + 
        amplitude.threshold
    ids <- data.frame(event = rep(NA, nrow(peaks)), landmark = NA)
    ids$event[not.noise] = detect.events(peaks[not.noise, var.columns], 
        var.weights)
    ids$landmark[not.noise] <- assign.to.landmarks(ids$event[not.noise], 
        n.landmarks)
    no.smooth <- -ncol(peaks)
    landmarks <- aggregate(peaks[not.noise, no.smooth], list(event = ids$event[not.noise]), 
        mean)
    names(landmarks)[-1] <- paste("mean", names(landmarks)[-1], 
        sep = ".")
    landmarks$duration <- aggregate(peaks$time[not.noise], list(event = ids$event[not.noise]), 
        function(x) {
            diff(range(x))
        })$x
    for (j in 1:n.landmarks) {
        f <- not.noise & ids$landmark == j
        if (any(f)) {
            temp <- aggregate(peaks[f, no.smooth], list(event = ids$event[f]), 
                mean)
            names(temp)[-1] <- paste(names(temp)[-1], j, sep = ".")
            landmarks <- cbind(landmarks, temp[match(landmarks$event, 
                temp$event), -1])
        }
    }
    landmarks <- landmarks[is.finite(apply(landmarks, 1, sum)), 
        ]
    ix <- grep("time", names(landmarks))[-1]
    landmarks[, ix] <- landmarks[, ix] - landmarks$mean.time
    answer <- list(file = basename(file.name), landmarks = landmarks)
    return(invisible(answer))
  }

landmarkacoustics/SoundPoints-R documentation built on Sept. 26, 2018, 9:08 a.m.