Audio segmentation using R, praat(?) and festvox.

List of R.Packages.

  1. Read wav file (sound)
  2. Cepstal analysis (tuneR? or read from festvox)
  3. Multivariate analysis of cepstral coefficients, pdfs of multivariate distributions (mvtnorm)
  4. Segmentation using turnpoints (pastecs)
  5. Multivariate means (colMeans)
  6. Clustering (stats, flexclust?)
  7. Evaluate clustering: cluster.stats in fpc

Idea: cluster intensity and/or deltas with Mclust. Here's an example: (:ipubfiles_maxwidth arctic_a0001_Mclust_intensity_classification.png:)

Intensity not scaled or standardized

GeSHi Error: GeSHi could not find the language r (using path /var/www/ (code 2)