Dolby Laboratories 585 - Manual (Page 38)

Model 585 Time Scaling Processor Principles of Operation
6-4
same characteristics (such as common onset, common harmonic properties, or in
multichannel audio, common spatial location), and breaks down the complete sound
into its component parts before discarding or adding samples.
Model 585 also takes advantage of psychoacoustic phenomena that have been studied
previously for use in audio coding schemes. For example, at smaller time scales, the
human brain focuses on a single auditory event at a given time. Therefore, processing
a relatively large section of audio allows multiple auditory events to be analyzed and
identified.
As the complexity of the audio increases, Model 585 identifies the most prominent
element at any given moment and adjusts the time scaling or pitch shifting
accordingly. It finds both the most prominent auditory events, as well as lesser
components that occur simultaneously. (The
Sensitivity setting controls the
degree to which lesser components are analyzed, as discussed in Section 2.6.2). Also,
by identifying the boundaries of events, which may exist in both the frequency and
time domains, crossfades between spliced sections can be performed with greatly
improved quality when compared to the methods that are used elsewhere.