Speech Enhancement And Noise Reduction In Forensic Applications
Surbhi Bharti, Prerna Jha, Medha Arora, Ashwni Kumar · 2023
This study explores the field of noise cancellation and speech enhancement in which relevant extraction of various audio features employing DNN(Deep Neural Netwrork) and MFCC (Mel-Frequency Cepstral Coefficients). MFCC method efficiently captures subtle spectrum characteristics by compacting audio streams into coefficients. The noise reduction and speech enhancement based data comprised on Deep Neural Networks (DNNs). DNNs learns to intricate mappings from noisy audio to clear speech using complex neural architectures. DNNs are capable of modelling non-linear interactions between noisy and clear voice inputs. DNNs are more successful than linear techniques Our study intends to harness the power of DNNs to successfully suppress noise, dramatically improving speech quality and expanding noise cancellation approaches through intensive training on various datasets. Forensic audio analysis is a crucial profession in criminal investigations and judicial proceedings, where audio evidence frequently acts as a lynchpin for revealing truth and enforcing justice. But background noise, distortion, and interference are typically present in forensic audio recordings, which can obfuscate important details and jeopardize the reliability of the evidence. This research study offers a thorough review of noise reduction and speech enhancement methods developed to meet the needs of forensic audio analysis which can be use as evidence.