JPS5919999A - Voice output unit - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a voice output device that outputs a specified input sentence in voice.

(2) Prior art and its problems In general, the voice output device shown in FIG. 1 is used to input a predetermined sentence manually and output the voice corresponding to the sentence. 2 in the figure, 1 is a switch Do.

Dl...-1・I) 7 is an input circuit for specifying and inputting a sentence to be output as voice, 2 is a microcomputer, 3 is a program old-sho'1.4 is a speech synthesizer, 5 is a sound μ data ROM, 6 is a speaker.

The conventional audio output device IT has an area M1 in the program ROM 6 for storing a control program for operating the microcomputer 2, as shown in FIG. 2(a).
, a storage area M2 for a conversion table that converts the input code that can be entered in the input circuit 1 into a sentence data address, and a sentence data storage area 13 that stores the number of phrases that make up the sentence and the address of the valley phrase for each valley sentence. and audio R
As shown in FIG. 2 (l), the OM5 stores two audio phrase data d1. The sound of this sentence is output from the speaker 6.

The operation of a conventional voice output device will now be explained using the control flow shown in FIG. The first j-cut 1 direct set is made, such as whether the wave is an impulse, triangular wave, or natural wave (Stetsubuko). Input code/sentence data address conversion table M24 in case of 1 person (Stellaf team)
Reference is made to convert the input code into a statement data address (step 3). The number of phrases in one sentence data corresponding to the sentence data address is read out from the sentence take storage area M (step 4), and then the first phrase address + a of the sentence data is read out (step 5).
, set the phrase address in the speech synthesizer 4 (
Step 6).

Next, send the voice start code to the speech synthesizer 4 (
Step 7).

Set Cyu1 utterance start code or set-g with speech synthesizer 4
Then, specify the address of the audio data ROM 5 based on the 7 race address specified in step 6,
Read the corresponding voice 7 race data, perform voice synthesis operation, and find out the 7th race? is output from the speaker 6.

In microlog 0 seso v2, it is determined whether or not this F# is being uttered (step 8).If 2 is being uttered, it waits at this step, but when 9 is finished uttering, the number of evening phrases is increased. If it is determined to be 0 (step 9), if it is not 0, proceed to step 5 and proceed to the operation of inserting the next statement data address.Same as above, from step 5 to step 8.
Processing in step 1 is line 7ffi, next 7 is the sound of lace? Deba. When it is determined in step 9 that there are remaining phrases or IJ, the designated person inputs the voice output of one sentence and ends the voice output, and returns to step 2 and waits at the next sentence.

Note that the remaining number of phrases in step 9 is confirmed by subtracting the number of phrases by 1 after each phrase is outputted and based on the result.

As mentioned above, with conventional voice output devices, to output one sentence aloud, several voice phrases are connected and output one after another, and the voice is not uttered naturally but in a smooth manner. It has the disadvantage of being lacking.

(3) Purpose of the Invention Therefore, the purpose of the present invention is to provide a speech output device that outputs speech sentences that are close to natural utterances.

(4) Structure and Effects of the Invention 1 Komei's sound blade device 1 that achieves the above objectives
indicates the sentence data 1. Number of phrases for each sentence.

sentence data storage means for storing addresses of each phrase making up the sentence and time data between each phrase;
voice data storage means for storing a plurality of voice phrase data respectively specified by each phrase address from the sentence data storage means;

a powerful means for specifying a sentence to be outputted aloud; and means for sequentially reading phrase addresses and time data of the corresponding sentence from the sentence data storage means in response to a sentence designation signal from the human means; , the voice phrase data specified by the read phrase address is forwarded to Ml”r.
Part 1: means for reading out data from the I data storage means, performing speech synthesis, and outputting speech for each phrase;
・The sound 7 that is output based on the current data at the time of departure
11 frame and a means for adjusting the length of time with a predetermined time width between the 11 frames and the Shindo frame.

According to the audio output device of this invention, in addition to each race I-' reply, the time data of each of the seven races m1 is stored as one sentence data, and the audio of one sentence is output (7) 2%
7 In response to <71. By setting this time data appropriately according to the preceding and following phrases, we can obtain sentence speech output that is very close to spontaneous speech with a pause between each speech phrase. be able to.

(5) Description of Examples The present invention will be explained in more detail with reference to two examples below.

The embodiment device described here is based on the audio volume J shown in FIG.
There is no difference from conventional audio output devices in that the basic configuration of the device is adopted. However, the data stored in the program ROMI and the functional configuration provided to the microcomputer 2 have been significantly improved.

FIG. 4(a) shows the program RO of the embodiment device of this invention.
It is a diagram showing the storage area arrangement of M3. An area M1 for storing a control program for the microprocessor 2 + a storage area M2 for a conversion gable that converts the input code input by the input circuit 1 into a statement data address
This device is the same as the conventional device in that it has a sentence data storage area M3 and a sentence data storage area M3. However, it is unique in that it has time data storage areas mtl, mt'2, and mL3 in addition to the phrase number storage area mn, phrase address storage areas ma1, ma2, ma3, and ma4 for each sentence data in the sentence data storage area M3. It is true. Note that FIG. 4(b) shows the audio key ROM5.
, and the manner in which the voice phrase data d1, d2, . . . , d8, . . . are stored is the same as before.

The apparatus according to the embodiment of the present invention operates under the control of the microcomputer 2 according to the control flow shown in FIG.

Next, the operation of this embodiment will be explained with reference to FIG. After the start of operation, after setting the initial value for the speech synthesizer 4 (step 11), the system waits in step 12 until input J circuit 1 receives a literate designation, but if there is an input in input circuit 1, step 12 The determination in step 1 is YES, and the storage area Mi is referred to to convert the human address 1-' into an exchange data address (step 13).

Assuming that the sentence data address converted here specifies the intersection data illustrated in FIG. 4(a), the phrase number 4 in the storage area Ill I+ is read (step 14), and then The phrase address stored in the storage area ma1, that is, the voice phrase data d2 is read (step 15).

Then, this phrase address is sent to the speech synthesizer 4.
(Step 16), and then a voice start code is set in the speech synthesizer 4 (Step 17). When this voice start code is issued, the voice synthesizer 4 addresses the voice data ROM 5 using the address of the voice phrase data d2 set previously, reads the voice frame data d2, and performs a voice synthesis operation. The speaker 6 outputs the voice of the phrase data d2.

While this phrase is being uttered, the control operation by the microcomputer 2 is on standby in step 18, but when 5 utterances are completed, 1 is subtracted from the number of read phrases in step 4 (at this point, the number of remaining phrases is minus 3). , the number of remaining shift phrases - 0 is determined (step 19). Since the first phrase is not 0 at the present time after it is uttered, time data of 23 mS is read from the fifth storage area mti (step 20). After waiting for 20 m5 (step 21).

Return to step 15. Then, the address of the next phrase address snawachi voice phrase data d4 is read. As in the case of the 1°f voice phrase data d2, steps 15 to 18 are performed.

This time, voice data ROM 5, voice phrase data d
4 is read out, synthesized by the speech synthesizer 4, and outputted as a sound from the speaker 6, when the utterance is completed.

Next, the time data 50m5 in the storage area m2 is read (step 20), and after waiting for 50m5, the process returns to step 15.

In this way, the processing operations from step 15 to step 21 are repeated until the utterance of all the huns of that one sentence is completed. During this period, the audio output is as follows: Voice phrase data d2 utterance → Time wait (201ns) → Voice phrase data d
4 utterance → time waiting (50mS) → voice phrase data d5
Voice → time wait (30 mS) → voice phrase data d7 voice.

In the above example, the number of phrases is 4, so the audio phrase data d
When the utterance of 7 is completed, the determination of the number of remaining shift phrases - 0 in step 19 becomes YES, and the process returns to step 12 υ
Prepare for the next literary talent.

If you set the time data properly in the above.

A suitable amount of time is maintained between the uttered phrases, making it possible to output more natural-sounding speech sentences.

[Brief explanation of drawings]

The first is a diagram showing the basic configuration of a general audio output device, and the second is a diagram showing the basic configuration of a general audio output device.
The figure shows the storage area arrangement of the ROM of a conventional audio output device, in which FIG. 2(a) shows the storage area arrangement of the program ROM, and FIG. 2(b) shows the storage area of the audio data ROM. FIG. 6 is a control flow diagram for explaining the operation of a conventional audio output device, and FIG. 4 is a diagram showing the storage area arrangement of the ROM of the audio output device according to the embodiment of the present invention. 4(a) is a diagram showing the storage area arrangement of the program ROM, FIG. 4(b) is a diagram showing the storage area arrangement of the audio data ROM, and FIG. 5 is an audio output diagram of the embodiment of the present invention. FIG. 3 is a control flow diagram for explaining the operation of the device. 1: Input circuit, 2: Microcomputer. 3 Niprogram ROM, 4: Speech synthesizer. 5: Audio data ROM, 6: Nupika. Ml: Control program storage area. M2: Input code/sentence data address conversion table area, M3 Two-sentence data storage area. m a 1 ・mn2...mn4: Phrase address storage area, +nti ・-mt2 ・mt3
: Time data storage area. Patent Applicant Tateishi Electric Co., Ltd. Agent Patent Attorney Shigeru Nakamura Nobu M1 Diagram

Claims (1)

[Claims] (1) A sentence data storage means for storing sentence data and storing the number of phrases for every nine sentences, the address of each phrase constituting the sentence, and the time take between each phrase; A voice take storage means for storing a plurality of voice phrase takes each specified by a phrase address, and an input means for specifying a sentence to be outputted as voice. Upon receiving the sudden designation input signal from this input means. means for sequentially reading out phrase addresses and time takes of corresponding sentences from said sentence data storage means for p times, and reading voice phrase data specified by the read phrase addresses from said voice data storage means to perform speech synthesis. A voice output device comprising: means for outputting voice for each phrase; and means for maintaining a predetermined time width between the output voice phrases based on the read time data.