Subjective audio quality impairments of playout adaptation with AAC RTP communication

Adaptive playout is a core component of almost every Internet-based real-time communication system. The main concept of playout adaptation is to schedule the playout time of incoming audio frames using time scaling techniques. The ACE (Audio Communication Engine), a low delay audio communication system performs adaptive playout by exploiting the structure of the underlying AAC codec. To the best of our knowledge, no in-depth study on the impact of these techniques on the audio quality has been accomplished so far. This paper presents the results of comprehensive subjective listening tests, assessing the quality of time stretching and shrinking mechanisms, as used by the ACE. The listening tests have been conducted according to the BS.1534 (MUSHRA) specification and cover different aspects of playout adaptation, for instance implicit re-buffering, frame dropping and combinatory effects. The test results clearly show the high audio quality of ACE´s time scaling techniques and point out their limits. In addition, the listening test results can be used as configuration guidelines to optimize the audio quality towards acceptable late loss and adaptation aggressiveness.