library(sound)

#----------------------------------------------------------------------
mysplit = function(d,nb.el.per.seg) {
   return(split(d, ceiling(seq_along(d)/nb.el.per.seg)))
}
#----------------------------------------------------------------------
next.T = function (splits, start) {
	for (i in (start:length(splits))) {
		if (splits[i] == T) { 
			return(i)
		}
	}
	return(NA)    # no more trues
}
#----------------------------------------------------------------------
# once the function is tested, create a similar function for next.F (next false)
next.F = function (splits, start) {
	for (i in (start:length(splits))) {
		if (splits[i] == F) { 
			return(i)
		}
	}
	return(NA)    # no more falses
}
#----------------------------------------------------------------------
create.word = function(split.signal, start, end) {
	return(unlist(split.signal[start:end]))
}
#----------------------------------------------------------------------
next.word = function(splits, above.silence, start) {
	signal.start = next.T(above.silence, start)
	if (is.na(signal.start)) {
		return(list(c(0), NA))
	}
	signal.end = next.F(above.silence, signal.start)
	if (is.na(signal.end)) {
		return(list(c(0), NA))
	}
	word = create.word(splits, signal.start, signal.end-1)
	cat("signal start/end ", signal.start, signal.end, '\n')
	return(list(word, signal.end))  # signal.end is next false
}
#----------------------------------------------------------------------
breakUpSoundWave = function(file, thresh, samples.per.segment) {
	snd=loadSample(file)
	snd = noSilence(snd,.0001)
	wave = snd$sound[1,]
	wave.abs = abs(wave)

	# if vol > thresh ,then "psychology"
	# if vol < thresh, then silence
	vol = wave.abs > thresh

	# split the wave into pieces of 1000 samples each
	# 1000 samples is about 0.05 seconds (1/20 sec). 
	splits = mysplit(wave.abs, samples.per.segment)
	
	# calculate the mean of every sample
	means = lapply(splits, mean)
	plot(1:length(means), means)
	above.silence  = means > thresh
	# print(above.silence)

	# redo the split to make sure that the words saved are not based on absolute value
	splits = mysplit(wave, samples.per.segment)
	#-------------------------------
  	sig.start = 1
	words = list()
	how.many.words = 0

  	while (T) {
    	ret = next.word(splits, above.silence, sig.start)
    	word = ret[[1]]
    	end = ret[[2]]
    	sig.start = end
    	if (is.na(end)) {
      		break  # exit loop
    	}
    	how.many.words = how.many.words + 1
    	words[[how.many.words]] = word
  	}
	return(words)
}

print("****")
#----------------------------------------------------------------------
#thresh = .001
#file = "psychology.wav"
#file = "assignment_sentence_instructor.wav"
#samples.per.seg = 3000
#words = breakUpSoundWave(file, thresh, samples.per.seg)
#----------------------------------------------------------------------