JPH0628898U - Voice message editing device - Google Patents

Abstract

(57) [Summary] [Purpose] Make it easy to correct the accent position. [Structure] The input Japanese sentence, for example, "Autumn trip was decided in Shimoda, Izu." / 00] Kimari Masita [.0
4] ”is created. The two-digit numerical value in [] is the information indicating the accent position, and based on this, the accent display conversion means 11 colors the syllable at the accent position in the phrase (accent mark), and the next display data. "a Kino [*] Ryokowa [,] is displayed in the size b Roh [/] Shimodani [/] Kimarimasu Ma Sita [.] is with the data display operation section 4 of. by a key operation of the coloring to look at the display The accent position is moved to correct the accent position (coloring is indicated by a line below the syllable), and the corrected display data is inversely converted into accent information (numerical value) by the inverse conversion means 12. And is converted into voice synthesis data.

Description

[Detailed description of the device]

[0001]

[Industrial applications]

 The present invention is applied to, for example, the creation of a voice message used in a voice response service, reads an input Japanese text, sets accents and inter-phrase coupling degrees, and converts it into voice synthesis data of a predetermined format. The present invention relates to a voice message editing device for displaying a voice message and correcting the voice synthesis data to edit the voice message, and particularly relates to the accent correction portion.

[0002]

[Prior art]

 Conventionally, when editing this kind of voice message, the accent information is corrected by a preset numerical value. For example, if the Japanese sentence "Autumn trip was decided in Shimoda, Izu.", The following voice synthesis data is created.

Aquino [* 01] Ryokoha [, 00] Izuno [/ 00] Shimodani [/ 00] Kimari Masita [.04] In this voice synthesis data, the kana string represents reading, and the symbol "*" in [] "," "/", ".", Etc. indicate the degree of connection between clauses, and the two-digit number in [] indicates the position from the beginning of the accented syllable.

[00] indicates a flat accent, that is, no accent. When correcting the accent, the operator grasps the accent position by converting it from a numerical value in advance or at any time.

[0004]

[Problems to be solved by the device]

 In this way, when the accent is corrected, the accent position (numerical value) is converted and grasped, so that it is not possible to directly know the position, and the accent information (numerical value) at the position that is not corrected is easily changed. There was a flaw. It is an object of the present invention to provide a voice message editing device capable of easily identifying the accent position and efficiently producing a high-quality voice message with no accent error in the voice message.

[0005]

[Means for Solving the Problems]

 According to this invention, based on the accent information of each bunsetsu, the accent display conversion means attaches an accent mark indicating the bunsetsu to the data for speech synthesis, and this accent mark is attached. The generated voice synthesis data is displayed as display data. Therefore, the accent position of each phrase can be read immediately. Also, the position of the accent mark is moved and corrected by keyboard input if necessary. The display data including the corrected accent mark or the uncorrected accent mark is inversely converted by the inverse conversion means by using the accent information (numerical value) corresponding to the phrase position.

[0006]

【Example】

 An embodiment of a voice message editing device according to the present invention is shown in FIG. When Japanese text including Kanji is input from the input terminal 1 to the text analysis unit 2, the text analysis unit 2 sets readings for Kanji, accents of phrases, and degree of connection between phrases to create data for speech synthesis. To do. The voice synthesis data is converted into display data by the voice synthesis data editing section 3 and sent to the data display / operation section 4. At this time, based on the accent information of each phrase, the accent display conversion means 11 adds an accent mark, for example, coloring to the accent position of the phrase. Therefore, in the data display / operation unit 4, the display data is displayed on the display unit, but the syllable corresponding to the accent position of each phrase is colored and displayed. Also, according to the operator's instruction, the coloring position is moved to change the accent position. The voice synthesis data editing unit 3 converts the data received from the data display / operation unit 4 from the display data format to the voice synthesis data format at any time. At that time, the inverse conversion unit 12 uses each accent mark, in this example. The colored position is converted back to accent information (numerical value). When receiving a command or the like from the data display / operation unit 4, the voice synthesis data editing unit 3 sends the edited voice synthesis data to the voice synthesis unit 5. The voice synthesizer 5 synthesizes a voice message 6 from the voice synthesis data.

The voice message editing apparatus according to the present invention performs voice synthesis data in a visually easy-to-understand form in order to facilitate the voice message editing by the operator. That is, the accent position is not represented by a numerical value that is difficult for a general user, but the position is displayed by coloring (accent mark) the kana character on the editing screen. Therefore, when the speech synthesis data created by the text analysis unit 2 has scattered accent errors, it is possible to efficiently obtain the desired speech message while repeating the correction and the generation of the synthesized speech. The above-mentioned correction process is as follows. For example, in response to the input Japanese sentence "Autumn trip was decided to Shimoda in Izu.", The data for speech synthesis automatically converted by the text analysis unit 2 is as follows: Aquino [* 01] Ryokowa [, 00 ] Izuno [/ 01] Shimodani [/ 00] Kimari Masita [.02] This voice synthesis data is converted into the next display data by the voice synthesis data editing section 3. It is colored and underlined below the syllable.

[0008] A Kino [*] Ryokowa [,] Lee's Roh [/] Shimodani [/] key Ma Rimashita [.] This display data is displayed on the display unit. In other words, the syllable "Aquino" is given a red coloring (accent mark), for example, as the accent position in the first syllable. In the syllable "Ryokowa", none of the syllables are colored and it is displayed as a flat accent. This display data is corrected by listening to the synthesized voice using the voice synthesis function. For example, “Izuno” sounds smoother with a flat accent, and “Kimarimashita” sounds more natural with a four accented “Ma”. The corrected display data is as follows.

[0009] A Kino [*] Ryokowa [,] Izuno [/] Shimodani [/] Kimarimasu Ma Sita [.] The display data is converted into voice synthesis data as follows. Aquino [* 01] Ryokowa [, 00] Izno [/ 00] Shimodani [/ 00] Kimari Masita [.04] Editing is performed by the operator referring to <display data>.

The key operation in the editing work can be easily performed by using, for example, a shift key or a cursor key. When the target voice message is confirmed, the voice synthesis data is saved and all the operations are completed. An operation method at the time of editing the accent position will be described. After editing, move the editing cursor to any position in the target phrase with the arrow keys, and then perform the following operations.

* Moving the coloring position to the left: pressing the "←" key while pressing the "SHIFT" key * Moving the coloring position to the right: pressing the "→" key while pressing the "SHIFT" key In the voice message editing device, in the voice synthesizing data editing process, V-RAM is used for the editing data in order to speed up the editing speed such as character input / insertion / deletion. In addition, in voice synthesis processing and paging processing, syntax correction processing is performed to determine whether the corrected voice synthesis data is in the correct format as voice synthesis data, and only normal V-RAM data is stored in the buffer. It is supposed to be stored. This plays a role in preventing rudimentary editing mistakes during editing.

Also, for displaying and coloring characters, a dedicated function created by the assembler is used, and sufficient editing processing speed is maintained to achieve the original purpose of this voice message editing device. It has become an effective means.

[0013]

As described above, according to the present invention, in the voice message editing device for editing the voice synthesis data, the information indicating the accent position is visually displayed instead of being displayed in a difficult numerical value. Since it is shown in an easy-to-understand manner, the accent position can be easily grasped, so that the desired voice message can be easily created.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a voice message editing device according to the present invention.

Claims (1)

[Scope of utility model registration request] 1. The input Japanese text is read, the accent and the degree of inter-phrase coupling are set, converted into voice synthesis data of a predetermined format and displayed, and the voice synthesis data is corrected. In a voice message editing device for editing a voice message, an accent display conversion means for attaching an accent mark indicating the accent position of the phrase to the voice synthesis data based on the accent information of the phrase, and the accent mark conversion means. A voice message comprising: an accent position correcting means for changing the position where the mark is made; and an inverse converting means for inversely converting the accent mark in the accented position-changed voice synthesis data into the accent information. Editing device.