Identification of Voice Segments in Stereo Soundtracks

While the human auditory system is easily able to distinguish between the voice and the audio components of a soundtrack, this appears to be a significant challenge to any computer model due to the overlap of audio and speech spectrum. The algorithm developed is able to discern the presence of voice in a stereo input with a high degree of accuracy. It is based on time-domain energy criterion, which results in lower computations in comparison to frequency domain techniques