**27.10.2016**

**Acoustics Research Institute**

9:30 - 10.15 Georg Tauböck (ARI) : *Phase Retrieval*

10:20 - 11:05 Martin Ehler (NuHAG): *Signal reconstruction from the magnitude of subspace components*11:30 - 11:55 Zdenek Prusa (ARI) :

*A Non-iterative Method for (Re)Construction of Phase from STFT Magnitude*

12:00 - 12:25 Peter Balazs (ARI) :

*12:30 - 12:55 Nicki Holighaus (ARI) :*

*The Pole Behaviour of the Phase Derivative of the Short-Time Fourier Transform*

*Reassignment*

14:30 - 15:15 Pejman Mowlaee (SPSC Lab) :

*Phase-Aware Signal Processing for Speech Communication*

15:20 - 15:55 Hisaaki Tabuchi / Bernhard Laback (ARI)

*:*

*Does the phase matter in hearing?*

**Abstracts:**

9:30 - 10.15 Georg Tauböck (ARI) : *Phase Retrieval We first give a brief introduction into the phase retrieval problem. Then, we present an information-theoretic framework for phase retrieval. Following an approach introduced for compressive sensing by Wu and Verdu (2010), we formulate phase retrieval as an analog source coding problem. Assuming that the signal to be recovered is random with a certain distribution, we derive novel achievability results. Our method is general in the sense that it holds for arbitrary distributions and includes sparse signals as a special case. *10:20 - 11:05 Martin Ehler (NuHAG):

*Signal reconstruction from the magnitude of subspace components*

11:05 -11:30

*Coffee Break*

11:30 - 11:55 Zdenek Prusa (ARI) :

*A Non-iterative Method for (Re)Construction of Phase from STFT Magnitude*

A non-iterative method for the construction of the Short-Time Fourier Transform (STFT) phase from the magnitude will be presented. The method is based on the direct relationship between the partial derivatives of the phase and the logarithm of the magnitude of the STFT with respect to the Gaussian window. Due to the non-iterative nature, the algorithm is very fast and it is suitable for long audio signals and it can also be easily adapted to the real-time setting. Moreover, solutions of iterative phase reconstruction algorithms can be improved considerably by initializing them with the phase estimate provided by the present algorithm.

12:00 - 12:25 Peter Balazs (ARI) :

A non-iterative method for the construction of the Short-Time Fourier Transform (STFT) phase from the magnitude will be presented. The method is based on the direct relationship between the partial derivatives of the phase and the logarithm of the magnitude of the STFT with respect to the Gaussian window. Due to the non-iterative nature, the algorithm is very fast and it is suitable for long audio signals and it can also be easily adapted to the real-time setting. Moreover, solutions of iterative phase reconstruction algorithms can be improved considerably by initializing them with the phase estimate provided by the present algorithm.

*The Pole Behaviour of the Phase Derivative of the Short-Time Fourier Transform*

*The short-time Fourier transform (STFT) is a time/frequency representation widely used in applications, for example in audio signal processing. Recently it has been shown that not only the amplitude, but also the phase of this representation can be successfully exploited for improved analysis and processing. In this talk we describe a rather peculiar pole phenomenon in the phase derivative, a recurring pattern that appears in a characteristic way in the neighborhood around any of the zeros of the STFT, a negative peak followed by a positive one. We describe this phenomenon numerically and provide a complete analytical explanation.*

12:30 - 12:55 Nicki Holighaus (ARI) :

*Reassignment*

13:00 - 14:30 Lunch Break

13:00 - 14:30 Lunch Break

14:30 - 15:15 Pejman Mowlaee (SPSC Lab) :

*Phase-Aware Signal Processing for Speech Communication*

The digital processing of speech signals is a key factor in human-machine interaction and is used in different everyday applications including assistive listening devices (digital hearing aids), mobile telephony, automatic speech recognition, speaker recognition, noise reduction and voice command devices. Since the invention of the telephone in 1876, large progress has been made towards advanced techniques developed to improve the achievable quality of the speech signal processing technologies experienced in everyday life. Among these efforts, the Fourier spectral magnitude based speech processing methods have been successfully employed while the spectral phase has been largely neglected.

In this talk, we first present an overview on the phase importance/unimportance beliefs in the literature. Then we focus on phase estimation fundamentals which is arguably required to initiate any phase-aware signal processing. In particular, reliable extracted information about phase can be utilized to push the performance achievable by the conventional speech processing methods that rely on the magnitude spectrum only. As some applications we exemplify speech enhancement, automatic speech recognition, and speech quality estimation (some demos are available at www.spsc.tugraz.at/PhaseLab). Finally, some conclusions and future directions will be provided.

The digital processing of speech signals is a key factor in human-machine interaction and is used in different everyday applications including assistive listening devices (digital hearing aids), mobile telephony, automatic speech recognition, speaker recognition, noise reduction and voice command devices. Since the invention of the telephone in 1876, large progress has been made towards advanced techniques developed to improve the achievable quality of the speech signal processing technologies experienced in everyday life. Among these efforts, the Fourier spectral magnitude based speech processing methods have been successfully employed while the spectral phase has been largely neglected.

In this talk, we first present an overview on the phase importance/unimportance beliefs in the literature. Then we focus on phase estimation fundamentals which is arguably required to initiate any phase-aware signal processing. In particular, reliable extracted information about phase can be utilized to push the performance achievable by the conventional speech processing methods that rely on the magnitude spectrum only. As some applications we exemplify speech enhancement, automatic speech recognition, and speech quality estimation (some demos are available at www.spsc.tugraz.at/PhaseLab). Finally, some conclusions and future directions will be provided.

15:20 - 15:55 Hisaaki Tabuchi / Bernhard Laback (ARI)* : **Does the phase matter in hearing? Phase sensitivity of the human auditory system has been investigated over the last seven decades. This talk reviews several basic monaural and binaural measures of phase sensitivity, including simple pure tones up to speech stimuli. *

*16:00 - 16:30 Coffee Break*

*Open Discussion *