Speech Processing Module for react.

• Record speech signals using computer microphone.
• Convert blob to wav files.
• Compute MFCC coefficients of signal.
• Training model for speech recognition.
• Define your own dictionary and train the application.
• Recognize spoken speech based on trained data.

The library mainly uses mfcc coefficients for storing the speech features. For recognition, it uses dynamic time warping using euclidean distance along with KNN modelling.
The library is best suited for the recognition of a smaller set of vocabulary with fairly distinguishing features across the trained vocabulary.

By default, the flow of the program includes first training letters A,B,C by providing 3 samples of each. Post that you can try recognition. You may also add the recognized features to the dataset for better accuracy.

Importing the library: import {SpeechProcessing, Rwcognize} from 'speech-processing-and-recognition-react';