JPS58158695A - Voice ornamentation system - 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 Object of the Invention The present invention relates to a voice modification method in a device equipped with a voice output function for the purpose of outputting voice and causing an operator to perform a desired action. The present invention relates to a voice modification method during repeated playback for listening back, which is particularly advantageous for devices using a voice generation mechanism based on a parameter synthesis method that tends to produce unnatural sounds.

Conventional technology Outputs information through speech synthesis and provides information to the operator.

For devices configured to perform manual operations,
The output audio information must be perceived and understood by the operator. However, in the case of a voice output device using voice parameter synthesis, the generated voice is generally unnatural. Conventionally, the sound has been repeatedly reproduced for this reason or for the omission of the sound, but this method is powerless against problems caused by the unnaturalness of the generated sound.

OBJECT OF THE INVENTION The object of the present invention is to provide an economical and effective repeat playback method when the operator forgets to hear or is unable to understand the sound in a device using an audio output mechanism based on the parameter synthesis method. It is about providing.

A predetermined sound pressure level is required for humans to clearly recognize speech. This value varies from person to person and also depends on environmental conditions (particularly noise level). It is a recognition target "
It also depends on the language. This may be the case when some words have similar sounds and it is not possible to determine which one is the correct one just by listening. In such cases, people
Trying to listen to the phonological nuances of words, we concentrate on the minute endings and beginnings of words.

Additionally, there are frequency bands that people can hear clearly, and although there are individual differences, female voices are generally easier to hear.

The present invention focuses on such human characteristics and attempts to improve the operator's recognition rate by modifying the voice in such a significant direction during repeated playback when listening and listening. Furthermore, by increasing the degree of modification each time it is played repeatedly (in other words, depending on the number of times it is missed),
The feature is that it can be approximated to the form that is easiest for the operator to identify. Specifically, in a storage mechanism that stores addresses of voice synthesis parameters, a storage section that stores the number of output repetitions of the parameters corresponding to each storage area in which the parameter addresses are stored is provided, and the number of repetitions is The system is configured to modify the voice in accordance with the above.

The storage unit that stores the repetition number may be located in correspondence with the parameter address and identified by a specific delimiter code, or it may be constructed by partially modifying the code itself that indicates the delimiter of the parameter address. good.

The memory part containing the number of repetitions is the corresponding parameter.

It is read out at the same time as the next reading, and a sound is generated.

It is sent to the department and used to modify the voice.

In order to modify the loudness of the generated sound by the number of repetitions, a variable gain amplifier is disposed immediately after the sound generation section, and the gain of the amplifier is controlled by the repetition count. In addition, in order to change the frequency band of the generated voice, the reference (pitch) frequency of the voice generation section is repeated in this manner. Therefore, the modification by the repetition number can be obtained by uniformly multiplying the pitch frequency specified by the parameter by the value determined by the repetition number. Ru.

Examples of the invention and its effects An embodiment of the present invention will be described below with reference to the drawings.

In the same figure, there is one information bit indicating whether the address is a parameter address in the depth direction of the storage mechanism where the parameter address is stored, and if this bit is OFF, the character position is used as a register to store the number of repetitions. This is an example in which a buffer configured to be used is used. In the figure, the method of writing the parameter address and the number of repetitions into the buffer is not explained. 1 is a control bit constituting one character of a buffer that stores a parameter address, and when this bit is 1, the character represents a parameter address. When this bit is 0, the character represents the number of repetitions of the immediately preceding or following parameter. 2 is the control bit 1 from the character in the buffer
The remaining information bits, excluding , indicate the number of repetitions or the parameter address. These information bits are respectively connected to registers 5 or 9 under the control of control bit 1 and are used as repeat counts or parameter addresses. The number of repetitions is received in register 6.

Connected to No. 4 D/A converter. The output of the vA converter 4 is connected to control a variable gain amplifier 5. On the other hand, the data in the nine registers divided as parameter addresses are connected to the parameter storage section 10, and the read audio parameters are stored in the parameter register 11. This parameter is connected to control the variable pulse generator 13 and the variable multistage digital filter 14 according to the meaning of the parameter. The output of the variable multi-stage digital filter 14 is sent to the D/A converter & low frequency door filter 1
5, the signal is converted into an analog signal and becomes an input to a variable gain amplifier 5 used as an audio amplifier. The output of the variable gain amplifier 5 is connected to a speaker 6. When control bit 1 is set to 1, the information bit indicates the parameter address, so it is stored in register 9, and the audio parameter with this value as the address is read out to parameter register 1, and the control value corresponding to the bit of the parameter is used. The variable pulse generator 13 and the multistage digital filter 4 are controlled to generate the desired sound.

On the other hand, the number of repetitions 2 written in the previous address buffer is read out to the register 3 under the control of the control bit 1. 4 is a DA converter which outputs an analog voltage according to the number of repetitions of the register 3; This analog voltage is applied to the voltage controlled variable gain amplifier 5 to increase the gain of the amplifier in accordance with the number of repetitions.

In this way, it is possible to obtain an audio output from the speaker 6 that increases according to the number of repetitions.

The dashed line in the figure is an example of a method for changing the standard pitch of output audio. Similar to the method of changing the audio output level, the number of repetitions is stored in the register 5, and a table 7 stores the multiplier of the reference pinch according to the number of repetitions.
given to.

The result is multiplied by register 8 and output.

The multiplication factor is given as one input of the multiplier 12.

On the other hand, the parameters read from parameter register 1 according to the parameter address of register 9, in this case, only the pitch control bit of variable pulse generator "13" is extracted and applied to one input of multiplier 120, The pulse generator 16 generates a pulse train based on the result of this multiplication, and as a result, the reference pitch is equal to the number of repetitions.

The speaker 6 obtains a sound that increases in pitch according to the amount of time. As described above, according to the present embodiment, the sound is sequentially amplified in accordance with the repetition count regarding the parameter stored in the buffer storing the parameter address. Alternatively, a voice modification method in which the pitch becomes higher (higher P) in sequence can be obtained with a simple configuration, and a significant effect can be obtained in playback when relistening.

Effects of the Invention According to the present invention, in a device that has a one-voice synthesis output function and forms an interaction with an operator, it is possible to significantly improve the recognition rate of the operator when listening to the operator's voice. A simple voice modification method is realized to realize repeated playback of .

[Brief explanation of drawings]

The figure is a block diagram of a device with an audio output function showing an embodiment of the present invention. Explanation of symbols 1... City U control bit, 2... Parameter address, 5... Register, 4... D/A converter,
5...Variable gain increase/heavy equipment, 6...Speaker, 7.
... table, 8 ... register, 9 ... register, 10 ... parameter storage section, 11 ... parameter register, 12 ... multiplier, 15 ... oJ variable pulse generator, 14 ... ...Variable multi-stage digital filter, 15.
...D/A converter & low range offshore wave device. Representative Patent Attorney Susuki 1) Yuki Toshi

Claims (1)

[Claims] A voice bar generated by a predetermined malprism. a voice generator including a memory mechanism for storing 2 meters, a voice synthesis unit for synthesizing sound and electric signals according to the parameters, and a control unit for instructing the voice generator to specify a parameter address of the voice to be generated; Predetermined. Parameters arranged to form a message. The audio output device is equipped with a second storage mechanism for storing a parameter address, and is characterized in that the number of times the parameter address is output from the second storage mechanism is counted, and the voice is modified according to the counted value. A voice modification method.