Loris for Your Cough

Roshan Mansinghani | Esmeralda Martinez | James McDougall | Travis McPhail

Home
Background
McAulay-Quatieri Method
Bandwidth Enhancement
Algorithm
   Loris Application

   Our Algorithm
Results
Conclusion
Poster
Group

Loris Application

After searching for applications to implement this analysis, we decided a software package named Loris was the best tool to use. This package was developed by the Cerl Sound Group (see bottom of page). Instructions for downloading and installing Loris follow:

 

Configuring Loris: For our project, we configured and installed the Loris software on a Redhat 7.3 linux box. First, the appropriate files must be downloaded:

Loris software (1.0a): http://sourceforge.net/projects/loris

FFTW libraries: www.fftw.org

SDIF libraries (at least 3.4): www.ircam.fr/sdif

Fossa software: http://sourceforge.net/projects/loris

The FFTW libraries implement a speedy Fast Fourier Transform (Fastest Fourier Transform in the West). The SDIF libraries (Sound Description Interchange Format) provide data formatting routines.

The FFTW and SDIF libraries must be compiled first. After uncompressing the archive they came in, a standard UNIX/Linux install took place:

 ./configure

make

make install

These compiled with no problems. Getting Loris to compile, though, was a bit tedious. After decompressing the archive, the same commands were run, but errors were given as we compiled (in the 'make install' stage). After much time, it was discovered that the install script was not placing the Python site packages in the appropriate directory. To minimize such errors, it is highly suggested that the user run './configure --help'. Here, a list will be printed to the screen where users can specify locations of libraries and target installation directories that our tailored to the individual's machine. After carefully reviewing the options, the standard UNIX/Linux install commands should be given along with the above-mentioned specifications.

After the Loris packaged has installed ('make') and the libraries installed to the appropriate directories ('make install'), it is very useful to run the test scripts available in the package.  Navigate to:

[loris directory]/test

Here, you will find all the files needed in order to perform a sound morphing routine given by Loris. There are several script languages that these routines are written in, but the Python script is the only one that is guaranteed by the producers, hence this is the one we used. Simply run:

python[or however you run python scripts] morphtest.py

There is some user feedback telling you what the script is currently doing. This script should take two aiff files in the test directory and morph them, creating other aiff files. It is very interesting to hear these results.

After ensuring that the Loris package has been successfully installed, the Fossa package can be next. After decompressing, the same install routine can be applied:

./configure

make

make install

After this, you should be able to run the program by typing 'fossa'. Here is a screenshot of the Fossa window: [look out for screenshot.png file in another email] Loris and Fossa are now ready to be used.