TRANSLATION DEVICE BASED ON VOICE RECOGNITION AND VOICE SYNTHESIS

摘要

A voice processing device, such as a translation device based on voice recognition. The voice is inputted through microphone (10) while operation switch is held down. Voice recognition section (14, 20, 22) recognizes the word spoken in a certain language which is contained in this voice signal by searching voice recognition dictionary (22), and outputs an indication which represents the recognized word. The recognized word is visualized on display device (32). If the displayed word is what is wanted by a user, the user waits for prescribed duration (t0) from this point of time. When prescribed duration (t0) elapses, voice synthesis section (16) reads, decodes and decompresses the voice signal from translation dictionary (36) where the voice signal for the word expressed in other language is stored in the compression-encoded form, and vocally outputs it from speaker (12). Through this procedure, the word expressed in the different language is outputted from speaker (12), corresponding to the voice inputted through microphone (10). These vocally expressed words may be visualized on display device (32) by operating key (44). If the word, which is displayed on display device (32) by voice recognition, is not what the user wants, the user can operates key (44) to interrupt the voice, which is being outputted by voice synthesis section (16) and output prescribed common phrases in the vocal form. The word to be inputted may be a key word, and in this case, voice synthesis section (16) outputs the sentences associated with the key word. Translation dictionary (36) take the form of storage medium (36) which is removably connected to the present device and is separately prepared for each of different languages and translation purposes. This removable storage medium (36) may contain renewable memory region (220a) for storing recognition dictionary (22). The voice signal of the user is renewably cataloged in recognition dictionary (22) as the reference voice signal in correspondence to the word. This cataloging is performed in the cataloging mode, but the cataloging is not performed, in response to the key operation by the user, for the word which does not require the cataloging of the reference voice signal. Further, translation dictionary (22) carries identification codes which shows the types and purposes of the words, and in cataloging mode, if the identification code held by control section (26, 34, 28, 30, 24) matches the code (124) in translation dictionary (22), the cataloging is not performed. The indication outputted from voice recognition section (20, 14, 22) is held (58) by control section (26, 34, 28, 30, 24), and if the user inputs the command to re-output it, control section (26, 34, 28, 30, 24) makes voice synthesis section (16) re-output the voice signal, which corresponds to the indication, based on this held indication (58).