The demand for high definition audio and video services is rapidly increasing. Two representative examples for this are audio-visual conferencing or video storage and delivery. In this context, efficient techniques are required for enhancement and compression of multi channel audio signals with compatibility to, e.g., mono or stereo systems. In this thesis, novel signal processing algorithms for both enhancement and compression of multi channel signals are developed and theoretical performance bounds are derived. Additionally, a novel instrumental quality measure for the evaluation of multi channel signal processing algorithms is proposed.Enhancement schemes for both the recording and the reproduction side are introduced. This includes the optimization of a near field filter-and-sum beamformer to achieve a target directivity characteristic at the recording side. For the reproduction side, an efficient postfilter is presented which increases the speech intelligibility by taking the positive influence of early room reflections into account.The main part of this thesis covers multi channel predictive compression of audio signals. A predictive multi channel coding system is presented and analyzed. Performance bounds are derived and two methods for an adaptive bit rate distribution between inter channel and intra channel prediction are devised. Novel multi channel noise shaping concepts are introduced. The performance of the compression system is quantified by instrumental measures.A novel instrumental measure is introduced for the evaluation of multi channel signal enhancement and compression. It combines the proven single channel quality measure PEAQ with a binaural auditory model and a mathematical model of cognitive behavior, providing a reliable evaluation of quality perception and spatial fidelity. The inclusion of spatial information into the instrumental quality measurement leads to a consistently high correlation between the instrumental measure and a listening test.

With the technological progress, devices, such as mobile phones, tablet computers or hearing aids, can be used in a large variety of every-day situations for mobile communication. Acoustic background noise signals, which are picked up with the desired speech signal, can impair the signal quality and the intelligibility of a conversation. A special noise type is generated outdoors, if the microphone is exposed to a wind stream resulting in strong-rumbling noise, which is highly non-stationary. As a result, conventional approaches for noise reduction fail in the case of noise induced by wind turbulences.This thesis is focused on the development of signal processing concepts, which reduce the undesired effects of wind noise. The key contributions are:• Signal analysis of wind noise• Digital signal model for wind noise generation• Signal processing algorithms for detection and reduction of wind noise signals.All these topics are considered with the focus on the development of algorithms for single and dual microphone systems.The analysis of recorded wind signals is the first step and gives valuable information for the estimation and reduction of wind noise. Furthermore it leads to a signal model for the generation of reproducible artificial wind noise signals.For the enhancement of the disturbed speech, an estimate of the underlying wind noise signal is required. In contrast to state-of-the-art noise estimation algorithms, the spectral shape and energy distribution is exploited for the distinction between speech and wind noise components leading to a novel estimation scheme of the wind noise short-term power spectrum. Considering a system with two microphone inputs, the complex coherence function of the two recorded signals is exploited for wind noise estimation. In addition to commonly used noise reduction schemes by spectral weighting, an innovative concept for speech enhancement is developed by using techniques known from artificial bandwidth extension. Highly disturbed speech parts are replaced by corresponding parts from an artificial speech signal.Objective measures indicate a significant increase of both the signal-to-noise ratio and the speech intelligibility. Besides, two application examples show that the proposed methods are very efficient and robust in realistic scenarios.

The audio bandwidth of digital landline and mobile telephone networks is still mostly restricted to 200 Hz to 3.4 kHz. This is due to compatibility requirements during the transition phase from analogue to digital transmission technology. The resulting characteristic "telephone speech" is widely accepted, but the intelligibility of syllables is only 91%.Meanwhile, improved coding standards for so-called “HD voice” or “Wideband Speech” have been developed which are gradually being introduced into the networks. They support an audio frequency bandwidth of 50 Hz to 7.0 kHz with significantly increased audio quality and speech intelligibility. For a very long time however, new HD-telephones and old narrowband telephones have to co-exist. If an HD-terminal is connected over a narrowband link to an old telephone, the improved coding scheme cannot be used.In this thesis, signal processing concepts are developed for improving audio quality and intelligibility of narrowband speech by artificial bandwidth extension (ABWE). These algorithms can be applied in the HD terminals or in the network, to transform narrowband speech to HD voice. Based on the source-filter model of speech production and a priori knowledge of the characteristics of speech signals, the missing frequency components between 3.4 kHz and 7 kHz are reconstructed. In comparison to the state-of-the-art ABWE approaches, the main contributions are:• new concepts of estimating the wideband spectral envelope, e.g., in terms of the model filter by interpolation in the acoustic tube domain• algorithms for spectral extension of the excitation signal• new insights concerning the relative importance of the excitation, the temporal envelope and the spectral envelope• remarkable improvements of the audio quality • significant low computational complexity• efficient and effective training and estimation algorithms The improvements are verified by objective evaluation and subjective listening tests.

The demand for high definition audio and video services is rapidly increasing. Two representative examples for this are audio-visual conferencing or video storage and delivery. In this context, efficient techniques are required for enhancement and compression of multi channel audio signals with compatibility to, e.g., mono or stereo systems. In this thesis, novel signal processing algorithms for both enhancement and compression of multi channel signals are developed and theoretical performance bounds are derived. Additionally, a novel instrumental quality measure for the evaluation of multi channel signal processing algorithms is proposed.Enhancement schemes for both the recording and the reproduction side are introduced. This includes the optimization of a near field filter-and-sum beamformer to achieve a target directivity characteristic at the recording side. For the reproduction side, an efficient postfilter is presented which increases the speech intelligibility by taking the positive influence of early room reflections into account.The main part of this thesis covers multi channel predictive compression of audio signals. A predictive multi channel coding system is presented and analyzed. Performance bounds are derived and two methods for an adaptive bit rate distribution between inter channel and intra channel prediction are devised. Novel multi channel noise shaping concepts are introduced. The performance of the compression system is quantified by instrumental measures.A novel instrumental measure is introduced for the evaluation of multi channel signal enhancement and compression. It combines the proven single channel quality measure PEAQ with a binaural auditory model and a mathematical model of cognitive behavior, providing a reliable evaluation of quality perception and spatial fidelity. The inclusion of spatial information into the instrumental quality measurement leads to a consistently high correlation between the instrumental measure and a listening test.

With the technological progress, devices, such as mobile phones, tablet computers or hearing aids, can be used in a large variety of every-day situations for mobile communication. Acoustic background noise signals, which are picked up with the desired speech signal, can impair the signal quality and the intelligibility of a conversation. A special noise type is generated outdoors, if the microphone is exposed to a wind stream resulting in strong-rumbling noise, which is highly non-stationary. As a result, conventional approaches for noise reduction fail in the case of noise induced by wind turbulences.This thesis is focused on the development of signal processing concepts, which reduce the undesired effects of wind noise. The key contributions are:• Signal analysis of wind noise• Digital signal model for wind noise generation• Signal processing algorithms for detection and reduction of wind noise signals.All these topics are considered with the focus on the development of algorithms for single and dual microphone systems.The analysis of recorded wind signals is the first step and gives valuable information for the estimation and reduction of wind noise. Furthermore it leads to a signal model for the generation of reproducible artificial wind noise signals.For the enhancement of the disturbed speech, an estimate of the underlying wind noise signal is required. In contrast to state-of-the-art noise estimation algorithms, the spectral shape and energy distribution is exploited for the distinction between speech and wind noise components leading to a novel estimation scheme of the wind noise short-term power spectrum. Considering a system with two microphone inputs, the complex coherence function of the two recorded signals is exploited for wind noise estimation. In addition to commonly used noise reduction schemes by spectral weighting, an innovative concept for speech enhancement is developed by using techniques known from artificial bandwidth extension. Highly disturbed speech parts are replaced by corresponding parts from an artificial speech signal.Objective measures indicate a significant increase of both the signal-to-noise ratio and the speech intelligibility. Besides, two application examples show that the proposed methods are very efficient and robust in realistic scenarios.

The audio bandwidth of digital landline and mobile telephone networks is still mostly restricted to 200 Hz to 3.4 kHz. This is due to compatibility requirements during the transition phase from analogue to digital transmission technology. The resulting characteristic "telephone speech" is widely accepted, but the intelligibility of syllables is only 91%.Meanwhile, improved coding standards for so-called “HD voice” or “Wideband Speech” have been developed which are gradually being introduced into the networks. They support an audio frequency bandwidth of 50 Hz to 7.0 kHz with significantly increased audio quality and speech intelligibility. For a very long time however, new HD-telephones and old narrowband telephones have to co-exist. If an HD-terminal is connected over a narrowband link to an old telephone, the improved coding scheme cannot be used.In this thesis, signal processing concepts are developed for improving audio quality and intelligibility of narrowband speech by artificial bandwidth extension (ABWE). These algorithms can be applied in the HD terminals or in the network, to transform narrowband speech to HD voice. Based on the source-filter model of speech production and a priori knowledge of the characteristics of speech signals, the missing frequency components between 3.4 kHz and 7 kHz are reconstructed. In comparison to the state-of-the-art ABWE approaches, the main contributions are:• new concepts of estimating the wideband spectral envelope, e.g., in terms of the model filter by interpolation in the acoustic tube domain• algorithms for spectral extension of the excitation signal• new insights concerning the relative importance of the excitation, the temporal envelope and the spectral envelope• remarkable improvements of the audio quality • significant low computational complexity• efficient and effective training and estimation algorithms The improvements are verified by objective evaluation and subjective listening tests.