A metric for determining the degree of compression of a processed signal
J.C. Schmidt, Janet C. Rutledge · The Journal of the Acoustical Society of America · 1998
When attempting to compare different compression algorithms or different parameter settings of a given compression algorithm, there is a need to compare equivalent amounts of compression. Listening tests that have varied only one parameter while fixing the others have obtained the expected results that increasing the amount of compression that occurs creates a more audible difference between the compressed version and the original. Since there are several parameters that influence the actual amount of compression that occurs, it is desirable to know which set of parameters will create to the least perceptual difference from the original signal for a given amount of compression. Unfortunately, there is not a well-defined approach for determining the amount of compression that has been imparted to a signal. The peak/RMS ratio is woefully inadequate for many compression contexts. Two degree of compression metrics are proposed that can be used to determine the amount of compression by direct analysis of the audio signal before and after compression. These metrics give excellent results in predicting the ratings of subject-based testing on the audio quality of several different audio segments across several parameter variations. a)Currently with ReSound. b)Currently with the University of Maryland Medical Center.