###############################################################################
###                     Version 0.2-0     Change-log                        ###
###                     May 13, 2013      <santiago barreda>                ###
###############################################################################

- Added new functions:

	+ combocalc() -> calculate combinations and permutations.
	+ findformants() -> locate the formants of a given sound using LPC.
	+ freqresponse() -> find the frequency response of a digital filter.
	+ lpc() -> perform LPC analysis. 
	+ loadtable() -> load a table without having to type a filename.
	+ ldaclassify() -> linear discriminant classification.
	+ ntypes() -> find the number of unique objects in a vector. 	
	+ peakfind() -> find the peaks in a vector.
	+ polezero() -> create a pole-zero plot from filter coefficients.
	+ pickIPA() -> select IPA symbols to be used for plotting.
	+ reduce.fraction() -> reduce a fraction to its lowest terms. 
	+ xsampatoIPA() -> convert x-sampa to IPA. 

	
- Added new vowel data sets:
	+ t07 -> Edmonton English vowels
	+ f99 -> Greek vowels
	+ p73 -> Dutch vowels
	+ a96 -> Hebrew vowels.
	+ y96 -> Korean vowels.
	+ b95 -> Spanish vowels.
	+ f73 -> Swedish vowels.
    
	
- Major changes:
	
	+ The addition of the pickIPA() and xsampatoIPA() functions allow for easy 
	  plotting of IPA symbols 'out of the box'. These functions use x-sampa to
	  translate an ascii representation of phones to the appropriate IPA 
	  symbols on the plot. 
	
	+ snipsound() is now just snip(), a general function to zoom in on, or 
	  'snip' subsections of spectrogram and sound objects. As a result 
	  spectrograms no longer have built-in zoom functionality.
	
	+ The resample() function has been made much faster by changing to sinc
      interpolation. As a result, resampling may now be by any rational factor. 
	
	+ The comparison methods of normalize.compare() have changed. The method
	  now calculates the percentage of correct classification using a linear 
	  discriminant classifier and the average between-category distance in the
	  (normalized) formant space. 
	  
	  	 
- Minor changes:
	
	  + vowelplot() can now plot using IPA symbols if provided with x-sampa 
	    representations of sounds.

	  + The vowel categories of all data sets are now represented in x-sampa. 
	    This allows for easy plotting of IPA symbols using vowelplot().

      + Spectralslice() now allows the creation of line spectra. 
	  
	  + The voice source for the vowelsynth() function has been changed to a 
	    simplified version of the KLGLOTT88 model. The amplitude envelope has 
		also been simplified. 
		
	  + The default color scheme for spectrogram() has changed slightly. The
        lowest color is now a darker blue, resulting in nicer spectrograms. 
		Another 'alternate'	color scheme has also been added.   
      
	  + The 'pause' parameter for spectrogram() is now FALSE by default.

	  + several functions have been sped up by changing from for loops to apply
	  (or a variant) wherever possible.

	  
- Bug fixes:
	  
	  + Fixed an error in the time alignment in spectrogram() causing plots to
	    be shifted in time by about 5 ms. 
	  
	  + Fixed the pooled, within-groups formant covariance matrix in PSTM().
	  
	  