JPS648840B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to a sound source circuit for providing sound source data corresponding to a sound source waveform to a digital filter (lattice filter). In general, in speech analysis and synthesis methods such as linear predictive coding, the resolution of the sound source is 8kHz.
(125μsec) or 10kHz (100μsec). Although this poses no problem in terms of sound quality for voice, the resolution is coarse for scales such as melodies and songs, which poses a problem in terms of sound quality. FIG. 1 shows an example of a conventional sound source waveform. Here, the "Γ" mark is the sampling point. The value of pitch is
As you can see from the figure, it has the resolution of the sampling frequency. In other words, if the sampling frequency is 8kHz, the value of the sound source will be in units of 125μsec. Taking each scale shown in Table 1 below as an example, the maximum error is 4.7%, and the permissible error for human auditory discrimination for scales is 1%, so speech synthesis for melodies and songs is There's a problem.

[Table] In order to eliminate the sound quality problems mentioned above, it may be possible to increase the pitch resolution of the sound source, but if the digital filter (sound source data calculation circuit) installed after the sound source circuit is 8kHz or
Due to restrictions such as those specified for 10kHz, it was not possible to increase the above resolution without modifying this digital filter. The present invention has been made in view of the above circumstances, and uses two or more waveforms of the same type that differ only in phase as sound sources in a speech synthesis circuit based on the analysis and synthesis method, thereby producing high-quality melodies and sounds using conventional digital filters. The purpose is to provide a sound source circuit that can obtain songs, etc. An embodiment of the present invention will be described below with reference to the drawings. FIG. 2 shows an example of the sound source waveform of the same embodiment, in which the sound source is a triangular wave. Figure 2a shows the basic waveform when the sampling frequency is 8kHz x 5. Figures b to f are obtained by setting the sampling frequency to 8 kHz and shifting the phase of the sound source waveform by 1/5 of the sampling period of 8 kHz with respect to the main waveform in figure a. FIG. 3 is a block diagram showing the configuration of the same embodiment, and _11b to _11f are ROMs (Read Only Memories) that store sound source waveform data. For example, the ROM 11 _b stores the data of the sound source waveform in FIG. 2 b (actually digital data at each sampling point indicated by the "Γ" mark), and the ROM 11 _c stores the data of the sound source waveform in FIG. (Actually, the digital data of each sampling point is indicated by an “×” mark)
is stored in the ROM 11 _f in the same manner as shown in Fig. 2 f.
Data of the sound source waveform (actually digital data of each sampling point indicated by "□") is stored. Reference numeral 12 denotes an address decoder for specifying addresses of the ROMs 11 ₁ to 11 _f . A counter 13 supplies address data to the address decoder 12, and the counter 13 is reset by a match output from the comparator 14. A data register 15 stores rough pitch data (integer) obtained from the analysis. Therefore, when pitch data n (integer), for example, is stored in this register, the counter 13 counts from "0" to "n" and supplies the count value to the address decoder 12 as address data. 16 is a data register that stores minute pitch data (0 to 4/5) obtained from analysis, and 17 is this register 16.
The multiplexer 17 selects the output of the ROMs _11b to _11f according to the data stored in the multiplexer 17, and the output of the multiplexer 17 becomes an input to a sound source data arithmetic circuit (digital filter). The input to this filter is therefore the ROM output selected by the data in data register 16. Fig. 4 shows the pitch period when the sound sources b to f in Fig. 2 are used. When the pitch value is n and the sound source waveforms used are b to f, the pitch values are n, n4. /5, n-3/5, 2-2/5, n-1/5
This means that the accuracy is five times higher than the conventional accuracy.
In terms of circuitry, whereas conventional sound source data was only the pitch value, the present invention requires the pitch value and data of the sound source waveform to be used, but between n and "n±1". Among the sound sources b to f located,
Because it is possible to select sound source waveform data that is closer to the sound source frequency to obtain the actually required scale frequency,
As shown in Table 2 below, the error for the musical scale is 1.2% at most, resulting in a significant improvement in sound quality.

[Table] Note that the present invention is not limited to the above examples,
Various applications are possible. . For example, when providing two or more waveforms of the same type that differ only in phase as sound sources,
Although ROM was used, sound source waveforms with different phases can be created by adding or subtracting arbitrary constants to the original waveform using an adder/subtractor. Further, in the above embodiment, a triangular wave is used as the sound source waveform, but the waveform is not limited to this, and various other waveforms such as a sawtooth wave and a trapezoidal wave can be considered. As explained above, according to the present invention, since a plurality of sound source waveforms with different phases are prepared and selected, the error for each scale can be suppressed within the range of permissible error for human auditory discrimination. Furthermore, it is possible to provide a sound source circuit which has the advantage of being able to obtain a high-quality melody using a conventional digital filter (lattice filter).

[Brief explanation of drawings]

Figure 1 shows an example of a conventional sound source waveform, Figure 2 shows an example of a conventional sound source waveform.
Figures a to f are diagrams showing examples of sound source waveforms according to an embodiment of the present invention, and Figure 3 is a block diagram showing the same configuration.
FIG. 4 is a sound source waveform diagram obtained by the same configuration. 11 _b to 11 _f ...ROM, 12... Address decoder, 13... Counter, 14... Comparator, 15, 16... Data register, 17... Multiplexer.

Claims (1)

Claims: 1. A sound source circuit comprising: data generation means for sound source waveforms that differ only in phase; and data selection means for each of the sound source waveforms. 2. Claim 1, wherein the phase shift between the sound source waveforms is equal to or less than the sampling period of the sound source waveforms.
The sound source circuit described in section. 3. The sound source circuit according to claim 1, wherein the output of the data selection means for the sound source circuit waveform is input to a digital filter for obtaining an audio waveform from the sound source waveform.