Comparison of Acoustic and Visual Voice Activity Detection for Noisy Speech Recognition

The problem of accurate differentiating between the speaker utterance and the noise parts in a speech signal is considered. The influence of utilizing a voice activity detection in speech signals on the accuracy of the automatic speech recognition (ASR) system is presented. The examined methods of voice activity detection are based on acoustic and visual modalities. The problem of detecting the voice activity in clean and noisy speech is considered. The speech signal was recorded in a real-life scenario in an office-like environment with the babble noise generated by the loudspeakers at different levels. The proposed method of visual voice activity detection is aimed at enhancing the accuracy of ASR when the ratio of signal to noise is low. The numerals in English language are used as speech material and Word Error Rate (WER) is employed for the evaluation purposes.