JPH05281984A - Method and device for synthesizing speech - Google Patents

Abstract

PURPOSE:To optionally vary the time required for speech synthesis and the quantity of the synthesized speech. CONSTITUTION:A KANJI(Chinese character)-KANA(Japanese syllabary) mixed character code sequence as an object of speech synthesis and degree information showing a degree N are inputted through an input part 1, a language processing part 3 generates a phoneme series and a rhythm information corresponding to the input character code sequence on the basis of a word dictionary 2, and a synthesis parameter generation part 5 takes the cepstrum parameters of a phoneme corresponding to the phoneme series out of a speech element piece file 4 by the degree indicated by the degree information to generate phoneme parameters and also generate rhythm parameters corresponding to the rhythm information. Those phoneme parameters and rhythm parameters are inputted to a speech synthesis part 6, which performs synthetic filtering of degree indicated by the degree information from the input part 1 according to the phoneme parameters and rhythm parameters to generate the synthesized speech.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice synthesizing method and apparatus for producing a synthetic voice from a character code string or prosody information and a phoneme sequence.

[0002]

2. Description of the Related Art Recently, various speech synthesizers have been developed which analyze a sentence containing kanji and kana and synthesize the speech information indicated by the sentence by a rule synthesis method and output the synthesized speech information. And this kind of speech synthesizer has begun to be widely used as a telephone introduction service in a banking business, a newspaper review system, a document reading device, and the like.

A voice synthesizing apparatus adopting this kind of rule synthesizing method basically puts a voice uttered by a human into a predetermined unit,
For example, for each CV (consonant, vowel), CVC (consonant, vowel, consonant), VCV (vowel, consonant, vowel), VC (vowel, consonant), LSP (line spectrum pair) analysis or cepstrum analysis is used. The phoneme information obtained by the analysis is registered in a phoneme file, and the phoneme file is referenced to generate phonetic parameters (phoneme parameters and prosodic parameters), and the sound source is generated based on these phonetic parameters. Is generated and synthetic filtering processing is performed to generate synthetic speech.

Conventionally, such a speech synthesizer requires dedicated hardware for real-time processing. The system configuration of this speech synthesizer is roughly divided into the following two types.

In the first configuration, a host computer such as a personal computer (PC) converts a kanji / kana mixed sentence into prosodic information and a phoneme sequence (language processing), and synthesizes synthesis parameters and sound sources with dedicated hardware. , Synthesis filtering, and D / A (digital / analog) conversion. On the other hand, in the second configuration, the dedicated hardware performs all the processes from the kanji / kana mixed sentence to the generation of voice. The dedicated hardware in any of the configurations has an LSI called DSP (digital signal processor), which has a high-speed product-sum operation, and a general-purpose M.
Most of them are composed of PU (microprocessor unit).

On the other hand, the processing capability of the personal computer (PC) or engineering work station (EWS) has been increased, and the D / A converter, the analog output section and the speaker have been installed as standard, so that the above-mentioned processing is performed. Can be done in real time with software.

In such a system, when there are few tasks being processed, there is no problem, but when there are many tasks, voice synthesis is often not performed in real time. Therefore, a silent section is inserted in the middle of the uttered word, which makes the voice very difficult to hear. This is because the time required for voice synthesis is constant, so even if you are working in real time with few tasks, as the number of tasks increases, CP will increase to other tasks.
It happens because the execution time of U is taken.

By the way, the voice quality of the voice generated by the voice synthesizer adopting the current rule synthesizing method is changed as follows: male / female / child / old man / old woman, utterance speed, voice pitch (basic pitch, Average pitch), stress level, etc., so that you can select the voice that suits your taste. However, those choices could change the voice quality, but not the quality itself.

At present, most of the synthesized speeches have a high degree of "clearness", which is familiar to a person who hears them for the first time, but is used to them. There is also a problem that the person who is listening tends to get tired when listening for a long time.

[0010]

As described above, in the above-described conventional speech synthesis technique, since the time required for speech synthesis is constant, speech synthesis can be performed in real time when the number of tasks is small. When there are many tasks, there is a problem that it cannot be performed in real time, and because the quality of synthesized speech is fixed, there is a problem that it is not suitable for long-term use.

The present invention has been made in consideration of such circumstances, and an object thereof is to arbitrarily change the time required for speech synthesis and the quality of synthesized speech by changing the order of synthesis filtering. A speech synthesis method and apparatus capable of performing

[0012]

A speech synthesis method and apparatus according to the present invention receives information indicating the order of phonological parameters or the quality of synthesized speech, and generates phonological parameters and prosody according to phonological sequences and prosodic information. It is characterized in that, based on a parameter, synthesis filtering of an order according to the input information is executed to generate a synthesized voice.

[0013]

In the above structure, the order of the phoneme parameter input to the synthesis filter is changed according to the order of the phoneme parameter or the information indicating the quality of the synthesized sound, and the synthesis filtering of the order is executed. As described above, according to the present invention, by changing the order of the phoneme parameter input to the synthesis filter, the amount of calculation in the filter can be increased or decreased.

Therefore, in a system in which a speech synthesis process including synthesis filtering is performed by a specific task process of a CPU that executes multitasking, when there are few other tasks to operate (or when the CPU has high capability).
It is possible to generate a synthetic voice in real time by selecting a high order for the above, and a low order when there are many other tasks that operate in reverse (or when the CPU has low capability). In addition, the quality of the synthesized voice can be arbitrarily changed by changing the order of the phoneme parameter input to the synthesis filter.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of the speech synthesizer according to the embodiment.

The voice synthesizing apparatus shown in FIG. 1 has a character code string including kanji and kana to be subjected to voice synthesis, and an input unit 1 for inputting control information of synthesized voice. This control information includes, for example, information (order information) for selectively designating the order N of the synthesis parameter to be input to the synthesis filter in the speech synthesis unit 6 described later.

The speech synthesizer shown in FIG. 1 is also input by the input unit 1 and a word dictionary 2 in which accent types, readings, part-of-speech information and the like of words and phrases to be speech-synthesized are registered in advance. The language processing unit 3 that analyzes the character code string using the word dictionary 2 and generates the corresponding phoneme sequence and prosody information.

The speech synthesizer shown in FIG. 1 also stores a cepstral parameter group obtained by previously analyzing the input speech for each arbitrary speech unit, and a speech unit in which information representing the degree of the cepstrum parameter is stored. It has a file 4 and a synthesis parameter generation unit 5 for generating phoneme parameters (here, phoneme cepstrum parameters) according to the phoneme sequence generated by the language processing unit 3 and the order information from the input unit 1. The synthesis parameter generation unit 5 also generates prosody parameters according to the prosody information generated by the language processing unit 3.

The speech synthesizer shown in FIG. 1 further performs sound source generation and synthesis filtering processing of order N based on the phoneme parameter generated by the synthesis parameter generation unit 5, its order information and the prosody parameter. And a speaker 7 for outputting voice
Have and. The D / A converter for converting the synthesized voice into an analog signal in the voice synthesis unit 6 is omitted.

The speech synthesizer having the above configuration is realized by a personal computer (PC) or an engineering work station (EWS) that executes multitasking, and has an input unit 1, a language processing unit 3, and a synthesis parameter generator. The unit 5 and the voice synthesizing unit 6 (internal sound source generation and filtering processing portion) are functional blocks implemented by the program processing of the CPU (execution of the voice synthesizing processing task).

Next, the overall operation of the speech synthesizer shown in FIG. 1 will be described. First, the input unit 1 inputs a character code string containing kanji and kana to be voice-synthesized and degree information indicating the degree N. The language processing unit 3 collates the character code string input by the input unit 1 with the word dictionary 2, and the accent type and reading of the word or phrase that is the target of the speech synthesis indicated by the input character code string. , The part-of-speech information is obtained, the accent type / boundary is determined according to the part-of-speech information, and conversion into a reading format of a kanji / kana mixed sentence is performed to generate a phonological sequence and prosody information.

The phoneme sequence and prosody information generated by the language processing unit 3 are given to the synthesis parameter generating unit 5. The order information input by the input unit 1 is also given to the synthesis parameter generation unit 5.

The synthesis parameter generation unit 5 extracts the cepstrum parameters of the phonemes corresponding to the phoneme sequence by the order N indicated by the order information given from the input unit 1 from the speech unit file 4 to generate the phoneme parameters. At the same time, the synthesis parameter generation unit 5 generates prosody parameters according to the prosody information.

The voice synthesis unit 6 is a synthesis parameter generation unit 5
The phoneme parameter and the prosody parameter generated by are input together with the order information given from the input unit 1 and temporarily stored. Then, the voice synthesizing unit 6 generates a synthesized voice represented by the input character code string by performing sound source generation and digital filtering processing in accordance with the input phoneme parameter and its order information and prosody parameter. An analog signal is converted by a D / A converter (not shown) and output to the speaker 7. In this way
A voice is generated from a sentence containing kanji and kana input by the input unit 1 and output from the speaker 7.

Next, detailed processing of the voice synthesis unit 6 of FIG. 1 will be described with reference to the flowchart of FIG. First, the speech synthesis unit 6 sets "1" in a counter variable "j" indicating a frame number, and sets [frame period] / [sampling period] in a counter variable "i" indicating the remaining number of samples to be processed per frame. = P is set as an initial value (steps S1 and S2). Here, the [sampling cycle] matches the cycle of the clock of the D / A converter (not shown).

Next, in accordance with the order information given from the input unit 1, the voice synthesis unit 6 selects the order indicated by the same information from the phoneme parameters and the prosodic parameters input from the synthesis parameter generation unit 5 and held. Phoneme parameters C0 to C for one frame (frame number is "j") corresponding to N
A synthetic parameter Rj consisting of N and a prosody parameter is selectively inputted (step S3).

Next, the speech synthesizer 6 uses the phoneme parameter C0.
Then, sound source data for one sample is generated (sound source generation) using the prosody parameter (step S4). Then, the speech synthesis unit 6 receives the generated sound source data for one sample and performs filtering (digital filtering) using the phoneme parameters C1 to C6 (step S5).

Upon completion of the filtering process of step S5, the voice synthesis unit 6 determines whether or not the order N indicated by the order information given from the input unit 1 is "6" (step S5).
6), when it is "6", the one sampling data (voice data) generated in step S5 is output (step S).
10).

On the other hand, when the order N is other than "6", the voice synthesizing section 6 executes filtering using the phoneme parameters C7 to C10 with the data generated in step S5 as an input (step S7). .. Then, the voice synthesizing unit 6 determines whether or not the order N indicated by the order information is "10" (step S8).

If the order N is "10" as a result of the determination in step S8, the voice synthesizer 6 jumps to the one sampling data output process in step S10. On the other hand, if the order N is other than “10”, the speech synthesis unit 6
Performs filtering of the phoneme parameters C11 to C20 using the data generated in step S7 as an input (step S9), and then proceeds to the one sampling data output process of step S10.

As described above, in this embodiment, when the order N indicated by the order information is "6", the filtering of C1 to C6 is performed, and when the order N is "10", the filtering of C1 to C10 is performed.
Otherwise, the filtering of C1 to C20 is executed.

When the one-sampling data output process of step S10 is completed, the voice synthesizing unit 6 subtracts "1" from the counter variable "i" (step S11), and determines whether this "i" is larger than "0". It is determined (step S12). If "i" is greater than "0", the speech synthesis unit 6 returns to the processing of step S4 and thereafter for the next one sample sound source generation and the order N filtering processing, and otherwise. That is, P sample (P =
When the processing of steps S4 to S12 for [frame period] / [sampling period]) is executed, 1 is added to the counter variable “j” indicating the frame number (step S13).

In this way, the voice synthesizing unit 6 executes the processing of steps S4 to S12 P times, and generates voice data for one frame (P samples). When voice data for one frame (P samples) is generated, that is, when the counter variable “i” is not larger than “0”, the voice synthesizing unit 6 sets the counter variable “j” to the number of frames to be voice synthesized. It is determined whether or not it is equal to or less than "F" (step S14), and if it is equal to or less than "F", the process returns to step S2 and subsequent steps to generate audio data for the next one frame. To finish.

In this way, the voice synthesizing unit 6 executes the processing of steps S2 to S14 only F times to generate voice data for F frames. In the flow chart of FIG. 2, C1 to C20 are all filtered except N = 6 and 10. However, in this embodiment, the order N that can be designated by the order information input by the input unit 1 is used. Is
It is limited to three of 6, 10, and 20, and the other orders are not specified.

In the speech synthesizer configured as described above, it is assumed that the order information indicating the order “20” (N = 20) is given to the input unit 1. Sampling period is 12
Assuming that the frame period is 5 μs and the frame period is 10 ms, FIG.
The value of P is "80". Also, in the speech unit file 4, the cepstrum parameter corresponding to each syllable is C0.
Up to C20 are stored.

The synthesis parameter generation unit 5 is a language processing unit 3
The cepstrum parameters C0 to C20 of the designated order corresponding to each phoneme of the phoneme sequence generated in step (4) are extracted from the speech unit file 4 and the prosody parameters are generated according to the prosody information. If the number of all frames F of the parameters obtained here is 500, the phoneme parameters are 21 × 500 = 10500, and the prosody parameters are 500.
It is an individual.

The voice synthesizing section 6 is a synthesis parameter generating section 5
10500 phoneme parameters and 5 generated by
Of the 00 prosodic parameters, the first one frame of phoneme parameters C0 to C20 and the synthesis parameter R1 consisting of prosodic parameters are input (step S3), and a sound source is generated based on the phoneme parameter C0 and the prosodic parameters (step S3). Step S4). Next, the voice synthesis unit 6 inputs the sound source data into the synthesis filter and executes filtering using the phoneme parameters C1 to C20 (steps S5 to S12). The voice synthesizer 6 performs the above step S
The processing of 4 to S12 is performed 80 times (for 80 samples).

After that, the voice synthesizing unit 6 inputs the synthesis parameter R2 for the next one frame (step S3) and performs the processing of steps 4 to 80 80 times. Then, the speech synthesis unit 6 performs these series of processes (steps S3 to S12) 500 times (for 500 frames). The voice data is output in step S10 during these processes.

That is, in the above example, the synthesis filtering using C1 to C20 is executed F × P = 500 × 80 = 4000 times. At this time, filtering 1 of C1 to C6
Filtering time T1 and C7 to C10
The time required for turning (steps S7 and S8) is T2, C11
Time required for one filtering of C20 (step S
If 9) is T3, and the time required for other processes in the series of processes shown in the flowchart of FIG. 2 is T4,
Speech time 5 seconds (frame 500 with frame period 10 ms)
Voice synthesizing unit 6 required to generate (for each) voice data
Total processing time at 4000 × (T1 + T2 + T3)
It becomes + T4.

Next, if the order N indicated by the order information is set to "6" under the same setting conditions as described above, the phoneme parameters extracted from the speech unit file 4 are C0 to C6, 7 × 5.
00 = 3500. Since N = 6, the voice synthesis unit 6
The processes of steps S7 to S9 in step S7 are not performed. The total processing time in this case is 4000 × T1 + T4, which is shortened by 4000 × (T2 + T3) as compared with the case of the order 20.

The cepstrum parameter generally has the property that the higher the order, the better the frequency spectrum envelope characteristic, and the lower the cepstrum parameter, the more the spectrum envelope curve becomes blunt. That is, the higher the order is, the higher the quality of the synthesized speech is generated, and the lower the order is, the lower the quality of the synthesized speech is generated. Therefore, by selecting the degree, the synthesized speech having the different quality can be generated. For example, when listening to synthetic speech for a long time, a low order may be selected.

As described above, according to the apparatus of this embodiment having the above-described processing function, the amount of calculation in the synthesis filtering can be increased or decreased by executing the filtering according to the order of the phoneme parameter. .. Further, it is possible to change the quality of synthesized speech by changing the order.

The present invention is not limited to the above embodiment. That is, in the embodiment, the case has been described in which one of the three orders of the cepstrum parameter determined in advance by the order information input from the input unit 1 is directly specified, but "1, 2, 3" or "A,
"B, C" etc. as the information indicating the quality of the synthesized voice,
It may be associated with the order of the phoneme parameter inside the device. Further, it is not necessary to limit the order that can be specified to three. Furthermore, the degree may be automatically set according to the CPU capacity or the CPU usage rate of another task (task other than the voice synthesis processing task), for example, CP.
The lower order may be set when the U capability is lower or the CPU usage rate of other tasks is higher.

Further, in the embodiment, the cepstrum parameter is used as the phoneme parameter, but other phoneme parameters such as LSP parameter may be used. In that case, a plurality of speech unit files corresponding to the analysis order are required.

Further, in the embodiment, the generation of synthetic parameters,
Although the case where the sound source generation, the synthesis filtering, and the like are performed in a system that is performed by software processing has been described, a system in which these processes are performed by dedicated hardware may be used, and the quality of the synthesized speech may be changed by changing the order. Can be changed. In short, the present invention can be variously modified and implemented without departing from the scope of the invention.

[0046]

As described above, according to the present invention, since the calculation amount of the synthesis filtering can be increased or decreased by changing the order of the phoneme parameter, the voice synthesis process including the synthesis filtering is performed by the specific task process of the CPU. In the system, voice synthesis can be performed in real time by setting the order according to the CPU capacity or the CPU usage rate of other tasks. Further, by changing the order of the phonological parameters, the quality of the synthesized speech can be arbitrarily changed, which has a great practical effect.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a voice synthesis device according to an embodiment of the present invention.

FIG. 2 is a flowchart for explaining the flow of processing of a voice synthesis unit 6 in the same embodiment.

[Explanation of symbols]

1 ... Input unit, 2 ... Word dictionary, 3 ... Language processing unit, 4 ... Speech unit file, 5 ... Synthesis parameter generation unit, 6 ... Speech synthesis unit, 7 ... Speaker.

Claims (2)

[Claims] 1. A voice synthesizing method for generating a corresponding phonological parameter according to a phonological sequence, a prosodic parameter according to prosodic information, and synthesizing a voice according to the phonological parameter and the prosodic parameter. A voice characterized in that information indicating quality is input, and based on the generated phoneme parameter and prosody parameter, synthesis filtering of an order according to the input information is executed to generate a synthesized voice. Synthesis method. 2. A voice synthesizing apparatus for generating a corresponding phonological parameter according to a phonological sequence, a prosodic parameter according to prosodic information, and synthesizing a voice according to these phonological parameters and prosodic parameters. Means for inputting information indicating quality, and speech synthesis for generating synthetic speech by executing synthesis filtering of an order according to the information input by the input means, based on the generated phoneme parameters and prosody parameters. A speech synthesizer comprising: