Custom Voice Spectrogram

The custom voice spectrogram program, developed for processing voice signals and improving voice quality, serves as the core focus of this work. This program enables comprehensive analysis and processing of voice data through key functions including signal preprocessing, Fast Fourier Transform (FFT) implementation, and mel-frequency cepstral coefficients (MFCC) extraction. The implementation involves frame blocking with overlapping windows, Hamming window application, and short-time Fourier transform (STFT) calculations to generate detailed time-frequency representations. For voice enhancement, the code incorporates spectral subtraction algorithms and Wiener filtering techniques to reduce noise while preserving speech characteristics. The program supports both real-time processing using buffer mechanisms and offline analysis with batch processing capabilities, providing improved clarity and intelligibility for various voice applications such as speech recognition systems and communication devices.