Skip to Main Content
The synthesis of speech by joining together segments derived from natural speech has not proven to be satisfactory with segments smaller than words, especially because of discontinuities in pitch and formant frequencies at the junctions. It appears that segmentation of the control signals for an analog synthesizer may avoid these difficulties. This paper describes an experimental system to investigate this method. A library of synthesizer control signals corresponding to subword segments of speech is now being developed. The equipment used to generate the library of control signals, as well as that used to synthesize connected speech from the library, is described. The synthesizer is a transistorized terminal analog of the cascade type. The synthesizer control signals are originally derived from functions drawn on a transparent plastic belt with opaque tape and scanned by a CRT and photomultiplier. The control signal functions are varied until the speech segment being studied is satisfactory. The resulting control signals corresponding to the speech segment are then automatically digitized and recorded on punched cards for addition to the library. Connected speech may be generated by computer assembly of the synthesizer control signals corresponding to a sequence of speech segments. In the assembly of connected speech from the library segments, pitch and timing may be specified independently of the sequence of segments if desired.
Note: The Institute of Electrical and Electronics Engineers, Incorporated is distributing this Article with permission of the International Business Machines Corporation (IBM) who is the exclusive owner. The recipient of this Article may not assign, sublicense, lease, rent or otherwise transfer, reproduce, prepare derivative works, publicly display or perform, or distribute the Article.