摘要 |
PURPOSE:To reduce a memory capacity to generate various kinds of sentence by providing a charge transfer memory, which stores fixedly prescribed sound information, and a means, which reproduces selectively storage contents from the memory above, and storing analogue information as it is. CONSTITUTION:Multilevel fixed charge transfer memory 1 constituted by charge transfer elements which store fixedly prescribed sound information is provided; and when a syllable is outputted from speaker 5, switches 2 and 6 are switched to memory 1, where the syllable is stored, by controller 8. After that, a fixed information read signal is given to memory 1 to shift information, which is generated in memory 1, by clocks which are supplied from clock generator 7 through switch 6, and the output is applied to LPF 3 through switch 2 to extract only sound signals by LPF 3, and these signals are amplified to output sound from speaker 5. Then, switches 2 and 6 are switched by controller 8 to combine syllables, so that various kinds of sentence can be generated. |