Quality assessment of speech signals under a process of echo cancelation in telecommunications systems

• Autorzy: Dobrucki Andrzej , Kin Maurycy , Walczyński Maciej

Identyfikatory

DOI

10.21008/j.0860-6897.2022.1.11

• Języki publikacji: EN

Abstrakty

EN

The phenomenon of echo in the telecommunications channels is caused by the reflection of an electrical signal in a long line. In order to improve the quality of the transmitted sound, various adaptive filters are used to remove or at least reduce the level of the reflected delayed signal. However, such a process may result in a degradation in the quality of speech, although its intelligibility may not get worse. The work presents the results of subjective studies on assessing the quality of speech signals under the process of acoustic echo cancellation using different algorithms. The algorithms studied were: LMS (Least Mean Squares), NLMS (Normalized Least Mean Squares) and AP (Affine Projection). The study consisted of assessing the signal quality after applying the echo elimination process using the Degradation Category Rating method. A total of 312 signals were used in the test: 192 male speech and 120 female speech samples. Echo simulation was used using different delay times and levels of echo signal. Both types of speech have signal delay times of 20 ms, 50 ms, 100 ms and 200 ms with echo level values of -6 dB, -12 dB, -18 dB and -24 dB. In addition, for female speech signals, a delay time of 150 ms was introduced. The study involved 14 people aged between 18 and 38, including six women and eight men. All subjects had normal good hearing. Seven listeners had participated in subjective listening tests of the sound quality assessment previously. The listeners’ opinions were collected on prepared questionnaire. It was found that the highest ratings were given to the AP filter, while the worst ratings were featured the NLMS. It should also be noted that the range between the results obtained for AP and NLMS for female speech is smaller in comparison to male. It is also interesting that the discrepancy in ratings was greatest for a delay time of 100 ms for the AP filter and 200 ms for the LMS filter. It can therefore be concluded from the obtained results that, in the case of acoustic echo cancelation, AP filter introduced the lowest quality degradation while the LMS achieved slightly worse average ratings when compared to the AP filter . The NLMS filter characterized by the worst ratings, and in some cases received twice the quality degradation compared to the AP filter.

Słowa kluczowe

EN

DSP; digital signal processing; echo cancellation; adaptive filtration; sound perception

PL

DSP; cyfrowe przetwarzanie sygnałów; kompensacja echa; filtracja adaptacyjna; percepcja dźwięku

• Wydawca: Poznan University of Technology. Institute of Applied Mechanics
• Czasopismo: Vibrations in Physical Systems
• Rocznik: 2022
• Tom: Vol. 33, nr 1
• Strony: art. no. 2022111
• Opis fizyczny: Bibliogr. 14 poz., 1 rys., wykr.

Twórcy

autor

Dobrucki Andrzej

• Department of Acoustics, Multimedia and Signal Processing, Wrocław University of Science and Technology

• https://orcid.org/0000-0002-2588-547X

autor

Kin Maurycy

• Department of Acoustics, Multimedia and Signal Processing, Wrocław University of Science and Technology

• https://orcid.org/0000-0001-5107-9327

autor

Walczyński Maciej

• Department of Acoustics, Multimedia and Signal Processing, Wrocław University of Science and Technology

• https://orcid.org/0000-0002-5126-7851

Bibliografia

• 1. H. Fastl, E. Zwicker; Psychoacoustics. Facts and Models; Springer, 2007. 2. A. Perry; Fundamentals of voice quality engineering in wireless networks; Cambridge University Press, 2007.
• 3. J. Benesty, T. Gänsler, D. R. Morgan, M.M. Sondhi, S. L. Gray; Advances in Network and Acoustic Echo Cancellation; Springer, 2001.
• 4. M. Walczyński, T. Zema; Model of active noise reduction system with adaptive filtering algorithms applied to automotive; In: Selected model based architectures and algorithms for learning, signal processing and optimization; Eds. Paweł Ksieniewicz, Mariusz Uchroński; Akademicka Oficyna Wydawnicza EXIT, Warsaw, 2021. s. 87-102.
• 5. D.T.M. Slock; On the convergence behavior of the LMS and the normalized LMS algorithms; Transactions on Signal Processing IEEE, 1993, 41(9), 2811-2825. DOI:10.1109/78.236504
• 6. T. Zema, M. Walczyński; Analiza zbieżności i skuteczności algorytmów filtracji adaptacyjnej w zastosowaniu do redukcji hałasu silników spalinowych stosowanych w samochodach osobowych (in polish); In: Nauka, badania i doniesienia naukowe 2019 : nauki techniczne i ścisłe, część I; Eds. T. Wysoczański; Idea Knowledge Future, Świebodzice, 2019, 353-368.
• 7. ITU-T: Recom, P.800, Method for subjective determination of transmission quality; Geneva, Switzerland, 1996.
• 8. EBU Technical Recommendation R22-1999; Listening Conditions for the Assessment of Sound Programme Material; EBU Geneva, Switzerland.
• 9. G. B. Kempster, B. R. Gerratt, K. Verdolini-Abbott, J. Barkmeier-Kraemer, R. E. Hillman; Consensus Auditory-Perceptual Evaluation of Voice; American Journal of Speech-Language Pathology, 2009, 18, 124-132.
• 10. S. Möller; Quality of Transmitted Speech for Humans and Machines; In: Communication Acoustics; Ed. J. Blauert; Springer, 2005.
• 11. A. Dobrucki, M. Walczyński, W. Bożejko; Parallel LMS-based adaptive algorithms of echo cancellation; In: Signal processing, algorithms, architectures, arrangements, and applications, Poznan, Chapters Signal Processing, Circuits and Systems, Poland Section; The Institute of Electrical and Electronics Engineers, 2014, 13-18.
• 12. A. Dobrucki, M. Walczyński, W. Bożejko; Family of parallel LMS-based adaptive algorithms of echo cancellation; Computational Methods in Science and Technology, 2015, 21(4), 191-200.
• 13. M. Walczyński; Acoustic echo cancellation in telecommunication networks using parallel computing methods (in polish); Reports of Department of Acoustics and Multimedia; Wrocław University of Science and Technology, 2014, PRE, no. 4.
• 14. A. M. Liberman, F. S. Cooper, D. P. Shankweiler, and M. Studdert-Kennedy; Perception of the speech code; Psychological Review, 1967, 74 (6), 431-461.

Uwagi

Opracowanie rekordu ze środków MEiN, umowa nr SONP/SP/546092/2022 w ramach programu "Społeczna odpowiedzialność nauki" - moduł: Popularyzacja nauki i promocja sportu (2022-2023).