SASPEEF – Statistical Analysis of SPEEch Features
Software SASPEEF is mainly designed for statistical analysis of speech features obtained from vowel records. If the user does not have separate vowels, SASpeeF can recognize vowel segments in fluent speech. From these segments, new sound files of each individual vowel are created for further analysis. The results of analysis (MFCC, LPC, LSP, fundamental frequency, formants etc.) can be compared with relevant previously saved results from database for finding the possible criterion for recognition the emotional state of speaker. If needed, current results can be saved into databse files for further processing. Obviously, the SASPEEF software can be helpful in the case of statistical processing of huge mined data from speech signals and for finding the most uniform behaviour and similarities in speech features varying on spoken vowel and speaker’s state.
signal processing, speech processing, vowels, formants, pitch, fundamental frequency, MFCC, LPC, LSP
UREL FEKT VUT v Brně