| Autor: |
Navya Amin, Markus Borschbach, Susanne Rosenthal, Marvin C. Offiah, Thomas Gross |
| Rok vydání: |
2015 |
| Předmět: |
|
| Zdroj: |
2015 International Conference on Smart Sensors and Systems (IC-SSS). |
| DOI: |
10.1109/smartsens.2015.7873594 |
| Popis: |
The requirements of a modern workplace as well as of socially integrated and autonomous aging demands a high quality of acoustic communication on the basis of interpersonal voice communication, perception and localization of warning signals, and generally acoustic information exchange. Increased listening effort results in a reduction of the concentration ability and performance with the final consequence of uncertainty and social retreat. A smartphone app-based hearing assistance system has been evolved that makes everyday acoustic scenarios more transparent by providing the opportunity to choose and focus on the preferred sound source which is enabled by the separation of actual signals in the existing acoustic situation into their respective signal sound sources. The main component of this assistance system is the blind source separation algorithm. This algorithm has to comply with several conditions regarding the compatibility with the system architecture, application problems and its separation ability. In this paper, a comparative study of four state-of-the-art Convolutive Blind Source Separation (C-BSS) algorithms is done on test audio files captured in four categories of acoustically challenging situations. This study does the comparison on the grounds of separation robustness and run-time efficiency. The best performing algorithm is further used as the separation technique in our hearing assistance system. The technical limitations of the state-of-the-art mobile systems for such an app and the possible alternates are suggested in this paper. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
