Digital Signal Processing Reference
In-Depth Information
Chapter 6
Audio Features
The ability to focus attention on important things is a defining
characteristic of intelligence.
—
Robert J. Shiller
To represent the information contained in the audio (stream) in a compact way
focussing on the task of interest, a parametrised form is usually chosen.These para-
meters describe properties of the audio usually in a highly information reduced form
and typically at a considerably lower rate, such as the mean energy or pitch over
a longer period of time. As different Intelligent audio analysis tasks are often best
represented by different such 'features', a broad selection of the most typical ones
will be presented in the ongoing—these will be the ones that are later also used in
the application examples in this topic. The determination of the features will include
the digitalisation and segmentation of the audio prior to their actual calculation or
extraction.
6.1 Audio Chunking
This section describes the digitalisation of audio and subsequent chunking in order
to go from an analogue stream to digitised chunks as 'units of analysis' that can be
processed computationally.
6.1.1 Digital Audio
In order to process the audio signal in a digital way, the analogue signal
s
ana
(
with
t
representing continuous time is represented by a sequence of equidistant (interval
t
)
t
)
with index
k
at the times
t
=
f
(
k
t
)
[
1
]. The area of these impulses is proportional to
