Chapter 3. Executable Programs

Table of Contents
General Information
Building your own executable programs
ch_wave Audio file manipulation
ch_track Track file manipulation
ch_lab Label file manipulation
tilt_analysis Produce tilt descriptions from F0 contours
tilt_synthesis Generate F0 contours from Tilt descriptions
sig2fv Generate signal processing coefficients from waveforms
spectgen Make spectrograms
sigfilter Filter waveforms
pda Pitch Detection Algorithm
pitchmark Find instants of glottal closure in Largynograph file
dp Perform dynamic programming on label sequences
ngram_build Train n-gram language model
ngram_test Test n-gram language model
viterbi Combine n-gram model and likelihoods to estimate posterior probabilities
na_play Audio Playback
na_record Audio file recording
wagon CART building program
ols Train linear regression model
ols_test Test linear regression model
wfst_build Build a weighted finite-state transducer
wfst_build Build a weighted finite-state transducer
wfst_run Run a weighted finite-state transducer
scfg_make Make the rules for a stochastic context free grammar
scfg_train Train the parameters of a stochastic context free grammar
scfg_parse Parse text using a pre-trained stochastic context free grammar
scfg_test Test the output of a parser
siod Scheme Interpreter
bcat Binary safe version of cat

General Information

Edinburgh Speech Tools provides a set of exectutables, which offer access to speech tools functionality in the form of a stand alone program. As far as possible, the programs follow a common format in terms of assumptions, defaults and processing paradigms.

The following are generally true of most speech tools programs.