Snack Home Page

The
Snack Sound Toolkit
for Ruby
(c) 2002 Stephen Legrand

Speech recognition, formant tracking, formant synthesis, dynamic time warping.

Introduction

Ruby is one of the hottest, newest languages which bundles features of from Java, Smalltalk, C++, and Eiffel (among others.) It is a perfect collaborative prototyping tool readily lending itself to basic research and the extreme programming environment. Thus marriage Ruby and Snack was inevitable.
The Snack Sound Toolkit is a collection of sound/voice processing routines used to extend scripting languages enabling any user to easily process speech. The original implementation inspired by Kåre Sjölander extended Tcl/Tk, and contains direct support for display of Spectrograms and Waveforms. Snack for Ruby is intended to provide these features from the ruby programming environment. Our initial implementation, tksnack, leverages the existing Tk-graphics by creating a ruby wrapper over the Tcl/Tk version of Snack. We hope that we will eventually extend this to other graphical environments. (Fox for instance.)

Screen Shots

The following are screen shots of sample applications.

SpeakEasy is a recorder featuring: recording, playback, echo effect, ...
Generator is waveform generator featuring: sin, square, sawtooth, noise, etc...
PowerDisplay shows the power distribution: recording, playback, etc...
SegmentationWorshop is simple segmentation program for separation sounds: recording, playback, etc...
Formant Filter applies formant filters to a given sound file.
Spectral Power Display allows the user to select a segment of speech and display the accumlated power spectrum for that segment.
FIR Display allows the user to do a quick inspection of the results of applying a FIR to a given signal.
IIR Display allows the user to do a quick inspection of the results of applying a IIR to a given signal.
Formant Display allows the user to select a segment of recorded speech and display the first and second formants for that segment.
Formant Monitor segments, monitors and displays the first and second formants of speech as it is spoken.
Formant Dynamic Time Warping stretches the feature sequences of a pair of sounds to align them with the best possible matching. This technique has been used to obtain a score the similarity of a pair of utterances. Demonstrates the use of dynamic programming.
Formant Archiver segments, displays, and saves formant feature sequences. It is very similar to Formant Monitor, except for its storage capabilites. It is used to create codebooks.
formantSynthesizer is a simple program to synthesize vowels .
whichWay is a simple Automated Speech Recognition program which points an arrow in the direction indicated by the speaker.

OnLine Help

For those with curiosity, on line help using the ruby-snack api is available here.

Installation

Snack for ruby (tksnack.rb) requires the snack package to be installed Snack requires Tcl/Tk to be installed. So depending on what you have on your system, there may be 1,2, 3 or 4 steps involved:
Unix Gurus

First download and install ruby. (if you don't already have it)
Second download and install Tcl/Tk (if you don't already have it)
Third download and install Snack (if you don't already have it)
Fourth download and install snack for ruby

Linux Groupies

First download and install ruby. (if you don't already have it)
Second download Tcl/Tk rpm and install Tcl/Tk(if you don't already have it)
Third download Snack rpm and install Snack(if you don't already have it)
Fourth download and install snack for ruby

Windows Addicts Precompiled binaries are available for windows users.

First download and install ruby. binaries (if you don't already have it) This will include tcl/tk.
Second download and install snack for ruby/win (rbsnack-tk/win) This will include a snack dll.

I confess, I haven't used the rpms, so I don't know if their status. If anyone does tell me:) Also I don't have access to a MAC, so if anyone wants to volunteer to help contact me:)

The Current State of Affairs

This site in the primordial stage of construction (and hence full of oversights and discrepancies) The current state of the code is best described as alpha.001 minus. Unit testing? (hah) We currently employ the Redmond model of QA.

Presentation at the International Ruby Conference 2002

Powerpoint presentation slides from the 2002 International Ruby Conference are available here.

Quote of the day

The goal of Computer Science is to build something that will last at least until we've finished building it.
- Anonymous

This website is hosted by SourceForge.net and VA Linux Systems. Thanks guys!

Last updated 30 October 2002