Subjective effects of variable delay and speech loss in dynamically managed voice systems
John G. Gruber, Leo Strawczynski · The Journal of the Acoustical Society of America · 1981
In this presentation we focus on speech impairments likely to arise in Dynamically Managed Voice Systems (DMVS). Here a DMVS is considered to be a system which may among other things use speech activity detection to exploit speech idle time and manage the resulting talkspurt traffic on a delay and on a loss basis. Examples of such systems include packet voice networks, buffered speech interpolators and advanced integrated services systems. Various speech quality classifications and test criteria used to evaluate the performance of speech systems are reviewed. Emphasis is placed on customer opinion of quality as defined by a rating scale, the objective being to provide the basis for exercising cost-performance tradeoffs. Based on recent subjective tests grade-of-service information is presented pertaining mainly to variable talkspurt delay and to various speech loss mechanisms including front-end and mid-talkspurt clipping. The importance of fixed delay in conjunction with echo, echo control and speech detection is also emphasized. Where possible comparisons with previous results are made.