You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
45 lines
1.1 KiB
45 lines
1.1 KiB
From text to speech: The MITalk system
|
|
|
|
TEXT
|
|
|
|
ANALYSIS
|
|
OMPONENTS
|
|
|
|
PHONO1
|
|
PHONOLOGICAL
|
|
COMPONENT
|
|
PROSODIC
|
|
COMPONENT
|
|
FOTARG PHONEMIC
|
|
SYNTHESIS
|
|
PHONETIC BY RULE
|
|
COMPONENT |: PHONET
|
|
STORED
|
|
PROSODICS
|
|
FORMANT SYNTHESIS
|
|
SYNTHESIZER
|
|
|
|
COEWAV
|
|
|
|
SPEECH
|
|
WAVEFORM
|
|
|
|
Figure 7-1: Synthesis blocks of the MITalk system
|
|
|
|
7.2 Background
|
|
|
|
Automatic voice response machines, based on the principle of concatenating
|
|
prerecorded speech waveforms, have been used to provide such information as
|
|
time of day and weather reports by telephone since the early 1930s. More
|
|
recently, voice response systems have been used to provide rapid telephone access
|
|
to information stored -in computers in such diverse areas as inventory control,
|
|
credit inquiries, bank balance information, and shipping status inquiries. In most
|
|
cases, the request can be keyed in by touch-tone telephone.
|
|
|
|
The earliest voice response units were analog systems in which the
|
|
vocabulary elements (words and short phrases) were stored as analog recordings of
|
|
speech waveforms. Many currently available audio response units still operate on
|
|
this principle (Homsby, 1972). Systems of this type have served very well in a
|
|
|
|
72
|