A set of research tools for silent speech classification, utilising KARA ONE database.
Download the repository, import Dataset and Classifier from SSC.py. I recommend taking a look into the example file. Both Dataset and Classifier are rather well documented.
- mne (https://martinos.org/mne/stable/index.html)
- Scipy
- Numpy
- Sklearn
- Matplotlib
- grid search goes ape on Windows if started outside "main" and there is parallel computing.
- finding best features works only with Anova and violates the assumption of normality of distribution.
- plotting
Wojciech Błądek
Project is licensed under the MIT License.
- Thanks to @stakar (Stanisław Karkosz) for a basis for the feature extraction module.
- Big thanks to the wonderful people behind KARA ONE (http://www.cs.toronto.edu/~complingweb/data/karaOne/karaOne.html)