Using a standard protocol and sample audio cases to enhance reproducibility, tests of coding quality are often performed jointly by laboratories around the world. Multiple Stimuli with Hidden Reference and Anchor (MUSHRA) is one such standard protocol. The same audio samples are used in all labs and as a result, listeners inevitably are judging quality in either their native language or one that they do not understand. It is not clear if a lack of understanding the language and its phonemes can influence the listener’s perception and his or her quality ratings during the test. This study used German and Mandarin Chinese speaking listeners, as well as test material in these two languages. The authors analyzed how ratings and listening times were affected by the foreign language. When results were pooled over all conditions, no significant differences between the ratings were found. However, for items of high audio quality, it was observed that listeners needed more time to evaluate samples that were not their native language, and it took more effort to compare different audio signals. As in MUSHRA tests – contrary to ITU-T P.800 tests - listeners can compensate for any difficulty they may have in perceiving artifacts by more effort and longer listening times, it seems to be no problem to include nonnative listeners in these tests at the expense of making them slightly less efficient.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.