I need to analyze about 30 seconds of an audio signal and detect if human voice is present. signal to noise ratio is good. I was wondering if there are any parts of speex that could perform this detection as opposed to traditional digital filtering.