作者: John G. Beerends
关键词: Sound quality 、 System under test 、 Speech recognition 、 Audio signal 、 Codec 、 Perception 、 Wideband 、 Computer vision 、 Computer science 、 Quality (business) 、 Artificial intelligence 、 Mean opinion score
摘要: A new, perceptual, approach to determine audio quality is discussed. The method does not characterize the system under test but characterizes perception of output signal system. By comparing degraded with ideal (reference), using a model human auditory system, predictions can be made about subjectively perceived any input signal. perceptual used calculate internal representations both and reference. simple cognitive interprets differences between representations. for assessment wideband music codecs as well telephone-band (300–3400 Hz) speech codecs. correlation subjective objective results above 0.9 wide variety databases derived from evaluations For measurement simplified given. This was standardized by International Telecommunication Union (Telecom sector) recommendation P.861.