Real Time FPGA Implementation for Speech Feature Extraction
Aditya Deshmukh, Arman Khan, Vedika Patil, Kishor Barasu Bhangale · 2025
Speech recognition is vital for many human-computer interaction systems and Mel frequency cepstrsl coefficients (MFCCs) plays vital role in feature extraction of the speech. This project aims to develop a Field Programmable Gate Array (FPGA) based application specific integrated circuit (ASIC) for extracting MFCCs from raw audio signals, and subsequently transmitting this feature data via USB to a laptop for further processing. MFCCs is widely used features in speech recognition, speaker identification, and other audio-based applications due to their ability to capture essential characteristics of the human voice. By extracting these features efficiently and transmitting them to a more powerful computing platform, this project seeks to provide a foundation for advanced speech processing tasks. The system will consist of an embedded platform (here, FPGA). It will be responsible for acquiring audio signals, performing preprocessing, extracting MFCC feature, and transmitting the feature data via USB to the laptop.