-
Improving Fairness in Speaker Recognition
Authors:
Gianni Fenu,
Giacomo Medda,
Mirko Marras,
Giacomo Meloni
Abstract:
The human voice conveys unique characteristics of an individual, making voice biometrics a key technology for verifying identities in various industries. Despite the impressive progress of speaker recognition systems in terms of accuracy, a number of ethical and legal concerns has been raised, specifically relating to the fairness of such systems. In this paper, we aim to explore the disparity in…
▽ More
The human voice conveys unique characteristics of an individual, making voice biometrics a key technology for verifying identities in various industries. Despite the impressive progress of speaker recognition systems in terms of accuracy, a number of ethical and legal concerns has been raised, specifically relating to the fairness of such systems. In this paper, we aim to explore the disparity in performance achieved by state-of-the-art deep speaker recognition systems, when different groups of individuals characterized by a common sensitive attribute (e.g., gender) are considered. In order to mitigate the unfairness we uncovered by means of an exploratory study, we investigate whether balancing the representation of the different groups of individuals in the training set can lead to a more equal treatment of these demographic groups. Experiments on two state-of-the-art neural architectures and a large-scale public dataset show that models trained with demographically-balanced training sets exhibit a fairer behavior on different groups, while still being accurate. Our study is expected to provide a solid basis for instilling beyond-accuracy objectives (e.g., fairness) in speaker recognition.
△ Less
Submitted 30 April, 2021; v1 submitted 28 April, 2021;
originally announced April 2021.
-
Coherent 100G Nonlinear Compensation with Single-Step Digital Backpropagation
Authors:
Marco Secondini,
Simon Rommel,
Francesco Fresi,
Enrico Forestieri,
Gianluca Meloni,
Luca Potì
Abstract:
Enhanced-SSFM digital backpropagation (DBP) is experimentally demonstrated and compared to conventional DBP. A 112 Gb/s PM-QPSK signal is transmitted over a 3200 km dispersion-unmanaged link. The intradyne coherent receiver includes single-step digital backpropagation based on the enhanced-SSFM algorithm. In comparison, conventional DBP requires twenty steps to achieve the same performance. An ana…
▽ More
Enhanced-SSFM digital backpropagation (DBP) is experimentally demonstrated and compared to conventional DBP. A 112 Gb/s PM-QPSK signal is transmitted over a 3200 km dispersion-unmanaged link. The intradyne coherent receiver includes single-step digital backpropagation based on the enhanced-SSFM algorithm. In comparison, conventional DBP requires twenty steps to achieve the same performance. An analysis of the computational complexity and structure of the two algorithms reveals that the overall complexity and power consumption of DBP are reduced by a factor of 16 with respect to a conventional implementation, while the computation time is reduced by a factor of 20. As a result, the proposed algorithm enables a practical and effective implementation of DBP in real-time optical receivers, with only a moderate increase of the computational complexity, power consumption, and latency with respect to a simple feed-forward equalizer for dispersion compensation.
△ Less
Submitted 9 June, 2015;
originally announced July 2015.
-
Optical Time-Frequency Packing: Principles, Design, Implementation, and Experimental Demonstration
Authors:
Marco Secondini,
Tommaso Foggi,
Francesco Fresi,
Gianluca Meloni,
Fabio Cavaliere,
Giulio Colavolpe,
Enrico Forestieri,
Luca Potì,
Roberto Sabella,
Giancarlo Prati
Abstract:
Time-frequency packing (TFP) transmission provides the highest achievable spectral efficiency with a constrained symbol alphabet and detector complexity. In this work, the application of the TFP technique to fiber-optic systems is investigated and experimentally demonstrated. The main theoretical aspects, design guidelines, and implementation issues are discussed, focusing on those aspects which a…
▽ More
Time-frequency packing (TFP) transmission provides the highest achievable spectral efficiency with a constrained symbol alphabet and detector complexity. In this work, the application of the TFP technique to fiber-optic systems is investigated and experimentally demonstrated. The main theoretical aspects, design guidelines, and implementation issues are discussed, focusing on those aspects which are peculiar to TFP systems. In particular, adaptive compensation of propagation impairments, matched filtering, and maximum a posteriori probability detection are obtained by a combination of a butterfly equalizer and four 8-state parallel Bahl-Cocke-Jelinek-Raviv (BCJR) detectors. A novel algorithm that ensures adaptive equalization, channel estimation, and a proper distribution of tasks between the equalizer and BCJR detectors is proposed. A set of irregular low-density parity-check codes with different rates is designed to operate at low error rates and approach the spectral efficiency limit achievable by TFP at different signal-to-noise ratios. An experimental demonstration of the designed system is finally provided with five dual-polarization QPSK-modulated optical carriers, densely packed in a 100 GHz bandwidth, employing a recirculating loop to test the performance of the system at different transmission distances.
△ Less
Submitted 9 June, 2015; v1 submitted 25 November, 2014;
originally announced November 2014.
