This is a repository copy of MetricGAN+/-: increasing robustness of noise reduction on unseen data.

This paper proposes neural models to predict Speech Intelligibility (SI),both by prediction of established SI metrics and of human speech recognition (HSR) on the 1st Clarity Prediction Challenge. Both intrusive and non-intrusive predictors for intrusive SI metrics are trained, then fine-tuned on the HSR ground truth. Results are reported on a number of SI metrics, and the model choice for the Clarity challenge submission is explained. Additionally, the relationship between the SI scores in the data and commonly used signal processing metrics which approximate SI are analysed, and some issues emerging from this relationship discussed. It is found that intrusive neural predictors of SI metrics when fine-tuned on the true HSR scores outperform the non neural challenge baseline.

Recent work in the domain of speech enhancement has explored the use of self-supervised speech representations to aid in the training of neural speech enhancement models. However, much of this work focuses on using the deepest or final outputs of self supervised speech representation models, rather than the earlier feature encodings. The use of self supervised representations in such a way is often not fully motivated. In this work it is shown that the distance between the feature encodings of clean and noisy speech correlate strongly with psychoacoustically motivated measures of speech quality and intelligibility, as well as with human Mean Opinion Score (MOS) ratings. Experiments using this distance as a loss function are performed and improved performance over the use of STFT spectrogram distance based loss as well as other common loss functions from speech enhancement literature is demonstrated using objective measures such as perceptual evaluation of speech quality (PESQ) and short-time objective intelligibility (STOI).

International and Comparative Law Quarterly [VOL. 44 charged with the application of competition law. His book puts forward a strong argument for a competition law based on fundamental rights of liberty and equality. Competition law is justified by the need for State regulation of the market to ensure that these principles are upheld. This leads Soriano to a historical analysis of administrative law drawing from the French and German experience. He points out that competition law can give way to other aims (p.50). Furthermore, Soriano has little time for economic arguments based on efficiency. He supports this approach with an account of an economist who defended corruption because it encourages the efficient exchange of information. In the author's view, there are principles which take precedence over such arguments of efficiency. Although Soriano draws support from the US approach (albeit "cum grano salts"), this rejection of any need to take account of economic theory of efficiency goes further than many anti-Chicagoans. The author completes his views on the principles of competition law with a strident call for privatisation and deregulation of Spanish industry, echoing the UK government's policies of the 1980s. Although Soriano rejects the simplistic labels of right and left (p.xxxi), this last point must involve the balancing of competing rights and aims-which surely involves a political decision. Do the views represented in Soriano's book reflect the attitude of the Tribunal? The first point is that the 1989 Royal Decree itself draws on fundamental rights protected by the Constitution. Moreover, decisions of the Tribunal seem to reflect Soriano's approach. In the important 1993 case, Liga de Futbol, the Tribunal fined the Spanish football league 147.5 million pesetas (about £700,000) for granting exclusive rights to a limited number of television stations to broadcast football matches on Spanish television. This decision seems more based on an idea of equality rather than any economic theory. The idea seems to be that all TV stations should be free to transmit the pictures even if this may reduce their value to the football clubs. Whatever one thinks about the approach described and advocated in Soriano's book, it raises interesting questions for comparative European competition law. The philosophy underlying a system of competition law is the most important guide to the application of the system. On this score, it may well be that the Spanish system is more open than many other systems. ROBBIE DOWNING East Centred European States and the European Communities: Legal Adaptation to the Market Economy. Edited by PETER-CHRISTIAN MCILLER-GRAFF (for the European Community Studies Association). [Baden-Baden: Nomos Verlagsgesellschaft. 1993. 235 pp-ISBN 3-7890-3135-6. No price given)

No abstract