Microphone Array Wiener Beamforming with emphasis on Reverberation

Detta är en Master-uppsats från Blekinge Tekniska Högskola/Sektionen för ingenjörsvetenskap

Författare: Vamsynagh Pedamallu; [2012]

Nyckelord: reverberation; wiener; beamforming; PESQ; SRP-PHAT;

Sammanfattning: Effective speech communication using microphone array is getting significant research in speech acquisition methods such as speaker localization, noise reduction and tracking. To locate the accurate Direction of Arrival (DOA) from the source, it is necessary to design a suitable microphone array system with noise reduction capability as well as efficiency in speaker localization. The goal of the system is to improve the quality of the primary speech signal. However, one of the major factors that affect quality of speech signal is the reverberation of the room. This reverberation, or reverb, is created when a sound is produced in an enclosed space causing a large number of echoes to build up and then slowly decay as the sound is absorbed by the walls and air. An Image method proposed by Allen and Berkley is used to determine the enclosure’s absorption coefficients in order to achieve the desired reverberation level. In this project the theoretical and practical use of image techniques for simulating on a digital computer and the impulse response between the two points in a rectangular room are discussed. The resulting impulse response when convolved with any desired input such as speech simulates the reverberation of the input signal. In addition this thesis describes the design and implementation of speech enhancement system in a single source environment with wiener beamforming. The proposed system consists of 4 microphones arranged in linear array. The setup is simulated in reverberant environment. Time domain and Filter-and-sum (frequency domain) beamformers are implemented. Weighted over Lap Add (WOLA) filter bank is also implemented as a part of frequency domain wiener beam forming in sub-band beamforming. As an extension SRP-PHAT algorithm is implemented and tested. Finally the quality of the speech is tested using SNR and PESQ (Perceptual Evaluation of Speech Quality) and also the performance of the system with respect to reverberation time is calculated. The results show that the two implementations are acceptable in terms of PESQ score.

  HÄR KAN DU HÄMTA UPPSATSEN I FULLTEXT. (följ länken till nästa sida)