Microphone front-ends for spatial sound analysis and synthesis with Directional Audio Coding
چکیده
Aalto University, P.O. Box 11000, FI-00076 Aalto www.aalto.fi Author Jukka Ahonen Name of the doctoral dissertation Microphone front-ends for spatial sound analysis and synthesis with Directional Audio Coding Publisher School of Electrical Engineering Unit Department of Signal Processing and Acoustics Series Aalto University publication series DOCTORAL DISSERTATIONS 33/2013 Field of research Acoustics and Audio Signal Processing Manuscript submitted 17 September 2012 Date of the defence 8 March 2013 Permission to publish granted (date) 14 December 2012 Language English Monograph Article dissertation (summary + original articles) Abstract A large number of professional and domestic audio applications utilize spatial sound reproduction. In addition to the conventional applications, such as the surround sound in movie and home theaters, spatial sound is also applied for telecommunication purposes. For instance in teleconferencing, sound emanated by talkers can be captured with multiple microphones at one end and reproduced spatially distributed with multiple loudspeakers at the other. This has benefit over a typical monophonic reproduction of the teleconference in terms of speech intelligibility and other elements of communication.A large number of professional and domestic audio applications utilize spatial sound reproduction. In addition to the conventional applications, such as the surround sound in movie and home theaters, spatial sound is also applied for telecommunication purposes. For instance in teleconferencing, sound emanated by talkers can be captured with multiple microphones at one end and reproduced spatially distributed with multiple loudspeakers at the other. This has benefit over a typical monophonic reproduction of the teleconference in terms of speech intelligibility and other elements of communication. During the last decade there has been an increasing research interest in parametric spatial sound processing. Several techniques for estimating the directional parameters of a sound field from multichannel audio files or from microphone signals have been proposed. In the parametric techniques, the directional information can be efficiently transmitted and then applied to spatial sound synthesis for various purposes. This thesis discusses Directional Audio Coding (DirAC) for capturing, transmitting and reproducing spatial sound. The perceptually motivated time-frequency processing of DirAC provides a parametric description of spatial sound, namely the arrival direction and diffuseness of sound. Direction and diffuseness, when analyzed in the time-frequency resolution of human hearing, are assumed to transmit enough information on the captured sound field for spatial hearing. DirAC has several applications of spatial audio, of which teleconferencing is mainly the focus here. The author's research addresses the development of different microphone front-ends for DirAC. The methods to analyze a sound field with input from arrays of omnidirectional microphones and from typical directional stereo microphones were studied. A novel method for diffuseness estimation was developed as a part of this work. Microphone arrays, which exploit an acoustic shadowing between microphones, are also proposed as an acoustical frontend for DirAC, as are the methods to conduct directional analysis with such arrays. These methods overcome the issues, which occur in direction analysis with input from the conventional microphone arrays, and thus provide reliable direction estimate over the entire audio frequency range. In the thesis, DirAC processing is also applied to bilaterally-fitted hearing aids with two microphones at each ear. The use of different microphone front-ends is evaluated through measurements and listening tests.