Phoneme identification from harmonics
[normalised amplitude harmonic 1]:

[normalised amplitude harmonic 2]:

[normalised amplitude harmonic 3]:

[normalised amplitude harmonic 4]:

[normalised amplitude harmonic 5]:

Predicted [phoneme] is:


(higher is more confident)
This prediction is the result of an algorithm using a Support Vector Machine. This algorithm has an expected accuracy of 85.81% on training data. However, there is absolutely no guaranty on the level of accuracy for predictions from data input by the user. The user of the prediction must ensure its accuracy using third party means before taking action or making decision based on the prediction. Any consequence of such action or decision is the entire responsibility of the user.
Dataset source: ELENA project, 1993
More information about this web app are available here.