Speech Enhancement using IOT for the Mute Community

Speech enhancement system focuses on improving people’s communication with speech impairments by converting unclear speech into a clear and intelligible output using a Rasp- berry Pi with a microphone. The speech is recorded using the microphone, processed for Mel-Frequency Cepstral Coefficient extraction and logistic regression reconstruction of classified speech features. The system works in real time and gives output through a speaker. Initializations had been done, including noise reduction and feature extraction for quality inputs to the logistic regression, which is modelled for indomitable setup through grid search with cross-validation. Testing was carried out in relation to effectiveness, which the system profiles through different metrics, especially accuracy, precision, recall, and F1-score. The results obtained are, in fact, that the model would be of help in increasing the clarity of the speech and hence providing quality in enhancing communication to real-time interactors with a speech defect. The system has the potential of impacting enhancing the quality of life for the user through enhanced communication.