JPS6059397A - Voice synthesizer - 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 (Technical Field) The present invention relates to a speech synthesizer that reads information in the waveform region of a phoneme waveform from a storage area and synthesizes speech, and particularly relates to a speech synthesizer that reads information in the waveform region of a phoneme waveform from a storage area and synthesizes speech.
Perform addition to superimpose PCM encoded phoneme waveforms,
This invention relates to a speech synthesizer that obtains speech output.

(Prior Art) When a human voice is a voiced sound, it is emitted when the airflow from the lungs becomes a quasi-periodic in-gurus flow through the vocal cords and resonates with a hollow resonator called the vocal tract.

Figure 1 shows the waveform of a human voice.

As shown in Figure 1, the human voice has almost the same waveform that repeats every cycle called ``pi'' and ``chi''.

This is because, as mentioned earlier, the human voice is emitted by a quasi-periodic impulse stream, and the pitch period is equal to the interval of this impulse stream.

When attempting to synthesize such voices, there are various methods depending on how the information N of the voice is stored.

One method is to store one-pitch period waveforms of speech (called phoneme waveforms) for various speech sounds in a storage area, and output speech by connecting these phoneme waveforms according to control information. be.

Figure 2 shows a configuration using this method.

1 is a storage area for storing phoneme waveforms with pitch periods for various voices; 2 is a storage area for control information such as pitch and amplitude multiplication for connecting phoneme waveforms;
3 is a synthesis unit that connects speech segments.

Note 1: Voices are synthesized by the synthesizer 3 by connecting the voices stored in the storage area 1 according to the control information stored in the storage area 2. However, in order to create a highly natural synthesized sound, the pitch of the voice must be adjusted. , the volume of the voice must be controlled in a continuous manner. The loudness of the voice is controlled by multiplying the phonemes in the storage area by a certain amount using amplitude multiplication information in the storage area. Furthermore, the pitch of the voice is controlled by pitch information in the storage area 2. However, the phoneme waveform in the storage area l has the pitch period length of the voice when the phoneme waveform was extracted, and does not necessarily match the pitch given by the control information.

Therefore, if the pitch of the control information is shorter than the phoneme length, the trailing end of the phoneme is cut off, and if the pitch of the control information is longer than the phoneme length, the last value of the phoneme is extended. Play phonemes with the same pinch length as the control information. (See Figure 3) However, if the phoneme waveform stored in storage area 1 is cut in the middle, if the phoneme waveform is not sufficiently attenuated, discontinuity will occur between the phoneme waveform and the connected phoneme waveform, which will affect the synthesized sound. On the other hand, when the phoneme waveform is lengthened, the wave tram of the phoneme waveform is deformed, resulting in deterioration of sound quality.

(Objective and Summary of the Invention) The object of the present invention is to solve these drawbacks, and to provide a speech synthesizer that superimposes I) PCM phase encoded phoneme waveforms of multiple channels shifted by pitch period. D
By time-multiplexing PCM code reproduction, and by sharing the adder used for the phoneme waveform superimposition calculation with the addition in the PCM code regenerator, natural, high-quality speech can be achieved with a simple circuit configuration. This is an implementation of a speech synthesizer, and will be explained in detail below.

(Premise of the invention) Previous: As mentioned above, sound is produced by the resonance of impulse flows from the vocal cords, and this can be replaced by an electric circuit.

In other words, speech is considered to be a superposition of output waveforms (hereinafter referred to as phoneme waveforms) of a resonant filter corresponding to one impulse generated by an excitation circuit corresponding to the vocal cords. This will be explained using FIG. 4.

2:'λ4 IO in the diagram is an in-wave train generated by the excitation circuit. Here, the time interval between each impulse is the pitch period interval. 11 is a synthesized audio output waveform obtained by driving a resonant filter with an in-Q pulse train generated by an excitation circuit.

12 is a phoneme waveform when the resonant filter is driven by the impulse P1 of the impulse train generated by the excitation circuit.

Similarly, 13 to 17 are phoneme waveforms when the resonance filter 4 is driven by impulses P2 to P6.

Since the driving impulse train 1° emitted from the excitation circuit is the addition of impulses P1 to P6, according to the superposition theorem, the synthesized speech output waveform 11 is obtained by the addition of each phoneme waveform 12 to 17.

The phoneme waveform 12 shown in FIG. 1 is O before time point t1.
It is. Time point 1. Thereafter, it has the property of attenuating with the passage of time and becoming 0 at an infinite time point. In fact, in the case of a speech waveform, the phoneme waveform is considered to be almost 0 when 16 milliJ seconds have passed from the excitation point (1+). Therefore, 16 milliseconds from the excitation time point (tl) is sufficient for the reproduction processing of the phoneme waveform 12.

However, in order to start the reproduction process of the phoneme waveform 13 from the excitation time point L2, multiple reproduction processes are required because the reproduction process of the phoneme waveform 12 has not yet been completed. The necessary waveform reproduction multiplicity n can be determined in the following step (1). In the case of normal voice, n of about 4 is sufficient. Therefore, the following explanation will be made assuming that n = 4.

Next, as a code for reproducing the phoneme waveform, a DI)CM code with high encoding efficiency is used.・For a regenerator that reproduces a waveform from a series of DPCM codes, Patent Application No. 55-09
address y of the quantization memory giving the quantization step size in the DPCM decoder proposed in 800;
The detailed explanation will be omitted since it is given by keeping the constant pointer value constant.

In a differential decoding process such as a DPCM code reproduction process, when the reproduction process is terminated after 16 milliseconds (128 sample periods), the final output value is held. Therefore, the phoneme waveform condition (Condition 1) is 0 before the excitation time point (Condition 2) and is 0 at the infinite time point.In order to satisfy the condition, the initial reproduction value must be O, and the reproduction final output value must be 0. It is necessary that the value is O, and even if the initial value for reproduction is O, the final value for normal reproduction will not be 0. Therefore, DPC is set so that the final playback value is 0.
It is necessary to provide M code strings.

(Example of the invention) FIG. 5 shows one embodiment of the present invention.

Here, the term ``1-channel'' or rchJ is used to number the processes that are time-divisionally processed within one sample period.

Each part in FIG. 5 is as follows. 100 is the first
c++ input register, 101 is 2nd channel input careno star,
102 is the 3rd channel input register, 1o3 is the 4th c1] input register, 104.105 is the selector, 106 is the DP
109 is a selector, 110 is a selector, 111 is a first . channel pointer register, 112, second channel pointer register, 113
114 is the 3rd channel pointer nosk, 114 is the 2nd channel pointer nostar, 115 is the selector, 117 is the quantization memory storing the quantization step value
0 is Renostar, 121 is Renostar that outputs synthesized voice,
122 is an output terminal, 135 is the 1st C11 count counter, 136 is the 2nd channel rerun counter, and 137 is the 3rd C
)] Down counter, 138 is the first Ich count, 139 is the second Ch decoder, 140 is the second Ch decoder, 141 is the third ch decoder, 142 is the fourth... ch
Decoder, 14 J is BIJSY signal of each channel, 1
'4.5 is a start signal for waveform reproduction, and 150 is a controller. The controller 15θ is the above-mentioned No. 101]
Input register 100, second C11 input register 101
+., 3rd channel decoder 141, 4th channel decoder 142, etc., and controls their operation, but the connections are omitted to avoid complicating the Figure 5 TD diagram. ing.

Figure 6 shows the data used in the present invention to reproduce the phoneme waveform, including the pointer value DPn and 128 DI'C11.
? Code Lnl J Ln2 + L12 + ”' r
It consists of Ln64. Further, the pointer value data DI'' is data stored in the pointer at the start of waveform reproduction.

FIG. 7 shows the relationship between the playback processing assigned to each skipped channel and time.

In FIG. 7, 200 is a synthesized output waveform, 20 is a phoneme waveform reproduced by the reproduction processing assigned to channel 1, 202 is a channel 1 activation signal ST1.203 is a channel 1 BUSY signal BUSY1. .

Below 205, 2 o 9 and 213 are channels 2 and 3
The phoneme waveforms 206, 210, and 214 reproduced by the reproduction process assigned to channel 2°3.4 are activation signals ST2. ST3. ST4.207,211゜2
1.5 is the BUSY signal BUSY2 for channels 2 and 3.4
.

BUSY3, BUSY4 signal is not working.

Embodiments of the present invention will be described in detail with reference to FIGS. 5 to 7.

At time point t1, a start signal STI is applied to the controller 150 from the outside and is transmitted through channel 1 as shown in FIG.
The reproduction process of the phoneme waveform corresponding to is started. At this time, the controller 150 sets the down counter 135 to a value of 128.
Set, 1. This value corresponds to the aforementioned 16 milliseconds. (If the sampling period of the output waveform is 125 microseconds, 16 milliseconds is 128 sampling periods.) If the output of the down counter 135 is other than 0, the decoder 139 outputs "1" as the BUS-11'l signal. Output.

Hereafter, data for waveform reproduction is read from the input current recorder 100 every sampling period (125 microseconds) and the DPCM
Perform playback processing. The down counter 1' 35 is decremented by 1 each time the reproduction process is completed. At the end of the 128th playback process, the down counter 135 becomes 0.9,B
The USYI signal becomes "o".

Table 1 shows the input current status (100,
101, 102, 1o3) DPCM code L stored in
n and the value shifted down by the shift register 118 (CXI, [Cause X], [IX), (,,x)
; However, CX) is performed in the l/ register 20 and the adder/subtractor 119 using the output of the quantization memory 117.
) PCM waveform re-key calculation processing (excluding pointer movement calculation).

Similarly to Table 1, at time point t2, a start signal ST2 is externally applied to the controller 150, and the reproduction process of the phoneme waveform corresponding to channel 2 (FIG. 4, 13) is started. At this time, the down counter 136 (1!, l; 2
g is set and the BUSY2 signal is 1''.Then, the same reproduction process as for channel 1 is performed, and after 128 sampling periods, the down counter 136 reaches O.The BUSY2 signal is ``1''.
o”.

From time point t3 onwards, in channel 3, time point 1,
Similar control is performed on channel 4 from time point t5 onwards, and on channel 1 from time point t5 onwards.

Playback processing using four channels like this? The synthesized audio output waveform is output every sampling period, and the reproduction processing of each channel is performed in a time-division manner at one sampling time point. Here, the time relationship of processing within one sampling period is shown in FIG. 8(a), and FIGS. 8(b), 8(c).

The processing within one sampling period is divided into five cycles, which are sequentially processed. It should be noted that most of the circuits required for the Dpcr4 raw processing processed in each cycle are shared, and all of the components shown in Figure 6, whose names include the channel number, are the same for each cycle. These are commonly used in 7
This is called the [shared part].

(Cycle j) The waveform reproduction process assigned to channel 1 is input to the current register 1θ0 + pointer register 111, down counter 135. This is done using the decoder 139 and a shared part,
The result is stored in register 120VC.

(Cycle 2) Reassigned to Channel 2/This waveform playback process is input! Interenosta 112. Down collar/
ri136. This is done using the decoder 140 and a shared part,
The result is stored in register 120.

(Cycle 3) The waveform playback process assigned to channel 3 is input to current screen 102. Pointerenosta 113. Down collar/ri 1
37, the processing is performed using the decoder 141 and the shared section, and the result is stored in the register 120.

(Cycle 4) The waveform reproduction processing assigned to channel 4 is performed by the input resouf 103. Pointerenosta 114. down counter 1
38. One step is performed using the decoder 142 and the shared section, and the result is stored in the register 120.

(Cycle 5) The value of the 7' star 120 is stored in the register 121 and used as PCM data for outputting synthesized speech.

Figure 8 (+) shows the ii control in the above five cycles.
).

This is shown in the flowchart in (c).

As described above, in this embodiment, the phoneme waveform reproduction processing using the DPCM code assigned to each channel is performed by time-division processing using the constituent elements for each channel and the commonly used shared section.

Furthermore, the sequential addition means (register 120 and adder/subtractor 119) used to reproduce the waveforms of each channel also serves as addition means for superimposing phoneme waveforms, and has the advantage of being realized by a small circuit configuration.

Furthermore, by changing the interval between activation signals of each channel, the pitch period of the synthesized audio output can be easily changed.

Furthermore, since one phoneme waveform also starts from O and ends at 0,
It also has the characteristic that sound quality deterioration such as waveform discontinuity due to the addition of individual phonemes is likely to occur.

(Effects of the Invention) As described above, according to the present invention, it is possible to configure a speech synthesis circuit that can control the pitch period and outputs high-quality synthesized speech with a simple circuit configuration.

[Brief explanation of drawings]

Figure 1 is a diagram showing speech waveforms, Figure 2 is a diagram showing the configuration of a conventional speech synthesizer, Figure 3 is a diagram showing waveforms used when changing pitch, and Figure 4 is a diagram showing the superposition of phoneme waveforms. Explanatory drawings, FIG. 5 is a diagram showing one embodiment of the present invention, FIG. 6 is a diagram showing a data format of a phoneme waveform, and FIG. 7 is a diagram showing a phoneme waveform and ST
A diagram showing the time relationship of the double signal BUSY signal, Fig. 8 (a
) is a diagram showing the time relationship of processing within one sample period time, and FIGS. 8(b) and 8(c) are flowcharts showing processing within one sample period time. 100-1st channel entry, 101...2nd channel
Input 1/Jista, 102 3rd channel input Kareno Star, 10
3 ・4th cl+l+input register 04・selector,
105 selector, lθ6 register, 109... selector, 110 selector, 111 i-th ch pointer register, 112 2nd cb pointer register, 11
3...3rd channel pointer register, 114/4th. c.
l + pointer register, 115... selector, 117
- ffi child memory, 11B... shift 1 register/star, 119... adder/subtractor, 12θ... register, 12 Luzo star, 122... output terminal, 1.95 1st c) 1 down counter, 136 2nd C1] Down count, 1
37. 3rd channel down counter, 138. 4th channel down counter, 139... 1st channel decoder, 140.
2nd channel decoder, 141... 3rd channel decoder 114
2. 4th channel decoder, 143-BUSY signal, 145
・Start signal, 150...controller. Patent applicant Oki Electric Industry Co., Ltd. Figure 6 Figure 8 (b' Figure 8 (c) Cycle ゛3! 1- Knife, l/4 Procedural amendment (voluntary) 1 Indication of case 1982 Patent application No. 1.67533 Name of No. 2 Speech Synthesizer 3 Relation to the case of the person making the amendment Patent applicant's office (105) 1-7-12 Toranomon, Minato-ku, Tokyo
No. 4 Agent address (105) 1-7 Kaori, Toranomon, Minato-ku, Tokyo
Contents of amendment No. 6 Attachment I) j- ('-5+'i-':), 9 6. Contents of amendment (1) Add one phoneme to line 10 on page 4 of the specification. ” is one phoneme. ” he corrected. (2) "DpcM@Unit 4" on page 7, line 10 of the same 11
A certain k r ADPCM decoder” is corrected. (3) Ministry of the same Ministry - 17Ln64'' on the 18th line of the 9th correction is corrected to rLnj'28j.

Claims (1)

[Claims] In correspondence with a plurality of channels, unit data including a pointer initial value and a plurality of DPCM codes is input to a reproduction means, and a phoneme waveform of each channel is reproduced to generate a phoneme waveform of each channel shifted by a pinch period. In a speech synthesizer that performs addition by an adding means and outputs the addition result as a synthesized sound, the DPCM code is a code set so that the initial value of one phoneme waveform matches the final value, and the DPCM code of each channel is The phoneme waveform is a waveform that has a fixed time length from the start of playback and has an initial value and a final value of 0. The early playback means is shared by each channel, and the playback processing in each channel is time-divided within the sample time. 1. A speech synthesizer according to claim 1, wherein said adding means shares a laser star and an adder/subtractor used in said reproducing means and accumulates phoneme waveforms.