R/ddiff.R

In emuR: Main Package of the EMU Speech Database Management System

##' Differentiation of tracks
##' 
##' Differentiates a list, as returned by track, to the nth order, readjusting
##' the index and ftime values each time.
##' 
##' 
##' @aliases ddiff ddiff.sub
##' @param dataset track data object - a list as returned by track
##' @param n the order of differentiation
##' @param smoothing if TRUE track is smoothed
##' @author Jonathan Harrington
##' @keywords math
##' @export ddiff
"ddiff"<- function(dataset, n = 1, smoothing = TRUE)
{
  ## differentiates a list, as returned by track, to the nth
  ## order, readjusting the index and ftime values each time
  ## dataset: a list as returned by track
  ## n: the order of differentiation
  
  ## now we apply the function to the data using dapply
  outdat <- dapply(dataset, ddiff.sub, n = n)
  if(smoothing)
    dsmooth(outdat)
  else outdat
}


##' @export 
ddiff.sub <- function(data, ftime, n)
{
  ## a function to be called by dapply
  ## data: a data matrix
  ## ftime: a start-end pair
  ## n: number of times to differentiate
  ## smoothing: if TRUE, apply smooth to data too
  ## returns: a list of $data values differentiated 
  ## and $ftime values adjusted accordingly
  ## values in $data that are returned are per millisecond
  if(is.matrix(data)) lval <- nrow(data) else lval <- length(data
  )
  if(lval < 1) stop("not enough data points in ddiff")	
  ## compute the time between samples
  interval <- (ftime[2] - ftime[1])/lval	
  ## do the differentiation
  data <- diff(data, differences = n)
  if(is.matrix(data))
    lval <- nrow(data)
  else lval <- length(data)
  timefactor <- (n * interval)/2
  ftime[1] <- ftime[1] + timefactor
  ftime[2] <- ftime[2] - timefactor	
  ## smooth the data as appropriate
  data <- data/interval	
  ## and return the data in the required format
  list(data = data, ftime = ftime)
}