JPS6339920B2 - - Google Patents

Description

[Detailed Description of the Invention] In general, a speech synthesizer is mainly composed of a word group storage unit and a speech generation control unit, and generates words or continuous word strings in response to instructions from an external device. . Normally, such words (sequences) are generated by assigning a code to each word (sequence) in the word group storage section, and using this code and an activation signal. FIG. 1 shows an example of the configuration of a conventional speech production system to which this speech synthesizer is applied. This generation system consists of a speech synthesizer control unit SCU and a speech synthesizer for uttering words (sequences).
It includes an SU, an amplifier AMP, and a speaker SP.
The speech synthesizer SU is composed of a word group storage section WS and a speech generation control section SG. In this device,
For example, to generate "HONJITSU", the following steps are taken. Here, in order to make the explanation more concrete, the explanation will be given from the stage of issuing a voice request to the system, although this is not directly related to the present invention. First, give a voice request SREQ to the system from outside the system.
This request is accepted by the speech synthesizer control unit SCU. The speech synthesizer control unit SCU recognizes that the speech request SREQ is "HONJITSU" from the request information. Next, the speech synthesizer control device SCU learns the voicing code SC for uttering "HONJITSU" and sends the voicing code SC corresponding to "HONJITSU" to the speech synthesizer SU through the bus B/ with the speech synthesizer SU. do. Furthermore, the speech synthesizer control device
When the SCU sends the activation signal ST to the speech synthesizer SU, the speech synthesizer SU performs, for example, based on the information for speech synthesis stored in the word group storage WS.
Synthesize the analog electrical signals of “HONJITSU”,
Send to amplifier AMP. The amplifier AMP amplifies this signal and converts it into a signal AS that enables electrical/acoustic conversion.
This signal AS drives the speaker SP, converts it into sound waves, and transmits them to human hearing through the air.
FIG. 2 shows an example of a series of timing charts for the various signals mentioned above. In FIG. 2, SI is a signal generated in the speech generation control unit SG and informing the speech synthesizer control unit SCU of the generation situation,
This signal is active during the period in which vocalization continues. Therefore, one cycle of development is the second point in time
This is the section from T ₁ to T ₂ .

By the way, if you say "WA" immediately after the above example, then immediately after that, say "SEITEN", and then again say "NARI", the speaking cycle given to the listener will be as shown in Figure 3. As can be seen from Figure 3, the vocalization becomes monotonous, giving the listener a sense of discomfort. For this reason, it is necessary to make speeches that are easy for listeners to understand. One possible way to do this is to lower the frequency of the clock pulses generated by the speech synthesizer SU shown in Figure 1, but in that case, the pitch of the utterance would be lowered, which would still create a sense of discomfort. Furthermore, it is also conceivable to provide an appropriate time interval between words, as shown in FIG. 4, for example. According to this method, the listener can understand the speech relatively well. However, this method does not have a lingering sound at the end, so it may be difficult for some people to hear it.

A first object of the present invention is to provide a speech generating device that can add lingering sound to the words being produced and increase redundancy to provide good speech to the listener. A second object of the present invention is to provide a speech generating device that can provide good speech according to words by variably adjusting the length of the added lingering sound.

The present invention will be described in detail below with reference to the drawings.

FIG. 5 shows an example of the configuration of the sound generating device of the present invention, in which the same parts as in FIG. 1 are given the same reference numerals. In Fig. 5, VS is a vowel storage device for generating vowels and exists as an extension of the word group storage device WS, and the speech synthesizer SU is a vowel storage device for generating vowels and exists as an extension of the word group storage device WS, a vowel storage device VS, and a speech generation device. It is composed of a control section SG. Speech synthesizer control device
The SCU consists of a speech synthesizer control main unit SCUM and an aftertaste table register ETR. This lingering sound table register ETR exists as an attribute corresponding to the utterance code similar to the example in Figure 1, and is a register that stores the lingering sound code table for uttering the lingering sound, and whether or not the lingering sound of a word (sequence) occurs. It is for storing the length of the aftertaste and the code for identifying the vowel corresponding to the word (sequence), and is made of, for example, a random access memory that can be operated from the outside.

Next, the operation of the sound generating device of the present invention shown in FIG. 5 will be described using as an example the case where "HONJITSUWAΔSEITENNARI" is uttered. A voice request SREQ is given to the system from outside the system in FIG. 5 in the same way as in the case in FIG. 1. The speech synthesizer control unit SCU sends a utterance code SC and a start signal ST to the speech synthesizer SU. At this time, the speech synthesizer SU sends an analog electrical signal corresponding to "HONJITSU" to the amplifier AMP, and the following is the same as the example in FIG.

By the way, while "HONJITSU" is being uttered, a request for the next utterance of "WA" reaches the system, but "WA" has not yet been uttered. On the other hand, the speech synthesizer control unit SCU is
While the speech synthesizer SU is uttering “HONJITSU”, the attributes of the voice code SC of “HONJITSU” are
Check whether a lingering effect occurs by referring to the lingering effect table register ETR. As a result, here, assuming that the voiced code SC "HONJITSU" has no attribute, "WA" will be generated immediately after "HONJITSU" is generated. “HONJITSU”
Similarly to the case of , for the next “WA”, the speech synthesizer control unit SCU registers the reverberation table register.
Examine ETR. As a result, if “WA” has an attribute, that is, it has a lingering sign, then it is a silent period (blank time) that has already reached the system.
Emit aftereffects before executing the request, then execute non-occurrence. In the same manner, "SEITEN" and "MARI" are uttered in sequence. FIG. 6 shows how such sound is generated.

In this example, the time to add "A" as a lingering sound after "WA", that is, one "A" is added, but the number of times "A" is added is controlled externally by the lingering sound table register ETR. It can be set to any desired number of times.

As described above, according to the present invention, it is possible to generate a lingering sound, and as a result, the sound field created by the equipment does not give a sense of discomfort to humans, and it is possible to create a very natural state, and to create an appropriate sound field. The user of the device can freely select the aftertaste. Moreover, by variably adjusting the length of the lingering sound, there is no need for a plurality of words (sequences) with different lengths of the lingering sound, so high performance/cost can be obtained without using expensive equipment.

Furthermore, with the voice generating device of the present invention, it is possible to utter unique lines by emphasizing the lingering sound.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of the configuration of a conventional voice generating device, FIG. 2 is a timing chart for explaining the operation of such a conventional voice generating system, and FIG.
FIG. 4 is a diagram illustrating the inconvenience caused when words or word strings are continuously generated; FIG. 4 is a diagram illustrating an example of utterance with no aftertaste; FIG. 5 is a block diagram illustrating an example of the configuration of the speech generator of the present invention; FIG. 6 is a timing chart showing an example of the sound generation operation according to the present invention. SU...Speech synthesizer, WS...Word group storage unit, VS
...Vowel memory unit, SG...Speech generation control unit, SCU
...speech synthesizer control device, SCUM...speech synthesizer control main unit, ETR...reverberation table register,
AMP...Amplifier, SP...Speaker, B1...Bass.

Claims (1)

[Scope of Claims] 1. In a speech generating device that inputs instruction information indicating a word to be generated from the outside and generates the instructed word by speech synthesis, the word is stored as information for speech synthesis. storage means; first information indicating whether or not a lingering sound should be added to the word corresponding to the word stored in the word storage means; and second information specifying a vowel to be added as the lingering sound. , a lingering memory means for reading out the first information and the second information according to an external instruction; a vowel memory means for storing third information for speech synthesizing the vowel used as the lingering sound; and a voice of the word specified by the instruction information. a speech synthesizer that performs synthesis; and a speech synthesizer that determines whether or not to add a lingering sound based on the first information, and if it is determined that a lingering sound should be added, a speech synthesizer that synthesizes the synthesized speech of the word and then adds the lingering sound to the second information. based on the third vowel corresponding to the last vowel of the synthesized speech.
A sound generating device characterized by comprising: a means for outputting information; and a means for outputting information. 2. In a speech generating device that inputs instruction information indicating a word to be generated from the outside and generates the instructed word by speech synthesis, the word storage means stores the word with information for speech synthesis, and the word. First information indicating whether or not a lingering sound should be added to the word corresponding to the word stored in the storage means, and second information specifying a vowel to be added as the lingering sound are stored, and the first information and A lingering memory means for reading out second information based on an external instruction; a vowel memory means for storing third information for speech synthesizing a vowel used as a lingering sound; and a speech synthesizing means for synthesizing a word instructed by the instruction information. Based on the first information, it is determined whether or not a lingering sound should be added, and if it is determined that a lingering sound should be added, following the synthesized speech of the word, the end of the synthesized speech is added based on the second information. 3rd corresponding to the tail vowel
means for outputting information; the lingering sound storage means stores time information regarding the time at which the lingering sound should be added, and based on the stored time information, the lingering sound is set as the lingering sound to be output to the speech synthesis means. A sound generating device characterized in that the number of times the third information is output is determined by a control means.