This system focuses on developing a speech recognition tool that compares spoken words to a reference text, evaluating both textual and pitch accuracy. Using the SpeechRecognition library, the system listens to and recognizes spoken words, converting the recognized text and the reference text into phonemes with the CMU Pronouncing Dictionary. The system utilizes the SequenceMatcher from difflib to assess textual similarity and librosa for pitch analysis. By comparing the median pitch and standard deviation of the recorded speech with reference values from a dataset, it adjusts the similarity score based on pitch accuracy. This approach allows the system to provide feedback on pronunciation accuracy, highlighting differences in phonemes and evaluating the overall correctness of the spoken input. The system also handles audio processing and temporarily saves recorded audio for analysis.
Article Details
Unique Paper ID: 164741
Publication Volume & Issue: Volume 10, Issue 12
Page(s): 2085 - 2088
Article Preview & Download
Share This Article
Join our RMS
Conference Alert
NCSEM 2024
National Conference on Sustainable Engineering and Management - 2024