JP3101668B2 - Memory integrated speech synthesis module - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital audio recording / reproducing device used as a man-machine interface between a human and a machine.

[0002]

2. Description of the Related Art Conventionally, digital voice information is stored as a memory element in advance, for example, in a memory LSI, and voice information is sequentially read out from the memory LSI, and voice synthesis is performed. Voice synthesis LS
I was present.

[0003] As a speech synthesis system, a "waveform encoding system" for digitally encoding and decoding an analog speech waveform, storing speech information as parameters such as amplitude and pitch as a dictionary and electrically generating organs. It is classified into an "analysis synthesis method" for simulating, and a "rule synthesis method" which has various speech parameters as a dictionary and handles arbitrary words.

In any of the above systems, two types of LSIs, for example, a memory LSI and a voice synthesis LSI, for storing various voice information and parameters are indispensable for voice synthesis output. FIG. 6 shows a component layout of a conventional speech synthesis module.

Conventionally, as shown in FIG. 6, for example, a memory LSI as the storage element 3 and a speech synthesis LS as the speech synthesis element 2
I are mounted side by side on the same printed circuit board 1 and are electrically connected to each other by pattern wiring. However, the audio information address transmission path 5 (address bus) and the audio information transmission path 4 (data bus Since the number of the pattern wirings is large, the wiring pattern mounting space 17 becomes very large, and there is a problem that the compactness and low cost of the voice synthesis module are hindered.

[0006] Further, in the case of writing audio information in the storage element 3 in FIG. 6, usually only once does not reach the completion area, the writing and audio reproduction of the audio information are repeated many times to check the audio information contents. It was necessary to repeat the recording until a satisfactory voice message content and sound quality were obtained. Therefore, a plurality of trials and errors are always repeated because the written contents are changed many times.

At this time, a rewritable EPROM (ultraviolet erasable programmable read only memory) is usually used as the storage element 3.
In this case, since an EPROM write control circuit is added on the printed circuit board 1, a new circuit mounting space is required, and there is a problem that the size of the voice synthesis module is further increased.

On the other hand, as another method, an external EPROM
When writing is performed using a writing device, a large number of external input / output pins, such as an EPROM address bus and a data bus, must be provided on the printed circuit board 1 for connection to an external device. Space is needed. For this reason, there is a problem that the size of the speech synthesis module is hindered from being reduced in size and cost.

When the voice information is fixed and completed as a voice synthesizing module after rewriting a plurality of times, the voice information in the storage element 3 can be easily read out, so that another person can copy the voice information to another storage element. By doing so, the same speech synthesis module can be easily duplicated. For this reason, there is a problem that it is impossible to keep valuable audio information confidential, and that copyright is easily infringed by others.

Therefore, as shown in FIG. 7, a speech synthesis LSI 18 with a memory in which a speech synthesis circuit section 20, a memory section 21 and a memory write control section 22 are integrated and mounted in one chip area 19 of the LSI has appeared. FIG. 7 is an internal configuration diagram of a conventional speech synthesis LSI with memory. Although this method can maintain confidentiality, the memory capacity is naturally limited due to physical restrictions on the chip size, and it has been difficult to increase the memory capacity at a practical level.

[0011] On the other hand, the demand in the industry for the speech synthesis module is increasing year by year. In particular, the two main demands are (1) a desire for higher sound quality and (2) a longer reproduction time, which is becoming the mainstream of needs on a practical level, and this tendency is becoming more remarkable with the years.

In order to achieve higher sound quality, it is necessary to increase the sampling frequency, and as a result, the amount of audio information increases. Similarly, in the case of longer-time reproduction, an increase in the amount of audio information is unavoidable, and in any case, due to a relative increase in the amount of audio information, it is becoming indispensable to increase the memory capacity. .

For the above reasons, the voice-synthesizing LSI 18 with memory shown in FIG. 7 has a physical defect that the memory capacity is insufficient at a practical level and cannot meet the needs of the present age. Also, in FIG.
In general, a non-erasable one-time PROM is built-in, and there is a limitation that once voice information is written, no correction or change of the voice information is possible.

As described above, when writing voice information into the memory unit 21, it usually does not reach the completion area by one time, and it is necessary to repeat trial and error several times until satisfactory voice message contents and sound quality are obtained. . For this reason, the speech synthesis LSI 18 with memory shown in FIG. 7 has a drawback that the device is discarded every time trial and error is repeated for addition, correction, and the like, resulting in economic waste.

[0015]

SUMMARY OF THE INVENTION The present invention solves the above-mentioned problems of the prior art as a voice synthesizing device, can cope with high-quality sound and long-time reproduction which is a trend of the times, and can rewrite voice information. It is an object of the present invention to provide a compact, low-cost memory-integrated speech synthesis module that can easily keep speech information confidential.

[0016]

In order to achieve the above object, the memory-integrated speech synthesizing module of the present invention has a large capacity as a storage element in order to realize high sound quality and long time reproduction, which is a trend of the times. A memory LSI is used.

Further, in order to facilitate rewriting of voice information,
Memory LSI via IC socket dedicated to memory LSI
Is implemented in the memory-integrated speech synthesis module. Also,
After the audio information is completely fixed, the audio information transmission path for reading the audio information is entirely sealed with a resin or the like in order to protect the copyright.

Further, before or after the audio information is fixed, the sealing is released by a special chemical or the like, and the digital audio information is stolen. Data conversion means for converting the read voice information into another digital voice information according to a special encryption rule.

Further, in order to realize a compact and low-cost module, the memory LSI and a voice synthesis LSI as a voice synthesis element are not provided on the same surface of the printed circuit board, but are provided on the front surface, for example, and the memory LSI is provided on the rear surface. Speech synthesis LSIs are arranged and mounted so as to overlap each other.

[0020]

The sound information is usually recorded by an announcer in a studio or the like, and then converted into digital sound information by a sound analysis tool, and then reproduced to confirm the sound. Next, the analyzed and edited digital audio information is written into a memory LSI by a memory writing device, and the memory LSI
Is implemented in the memory-integrated speech synthesis module.

Further, the memory-integrated speech synthesis module is mounted on a printed circuit board of various speech synthesis output devices. The speech synthesis LSI reads out speech information from the memory LSI, synthesizes the speech information, and then converts the speech information into an analog speech signal. Convert to waveform and output. The output is amplified by an amplifier circuit mounted on a printed circuit board of the voice synthesis output device, and then reproduced as voice by a speaker, earphone, or the like.

In order to improve sound quality and enable long-time reproduction, more and more audio information must be stored in a large amount. Therefore, a large amount of audio information is written and stored using a large-capacity memory LSI. Voice information in memory L
If a memory write control circuit and an input / output connector for connection to a memory writing device are provided in the memory-integrated speech synthesis module for writing to the SI, a very large mounting space is required.
When the memory LSI is mounted via the C socket, no extra space is required. The memory LSI can be attached and detached with one touch, and the memory LSI can be set in an external memory writing device to easily rewrite audio information.
Since the change processing can be performed, it is possible to contribute to downsizing and cost reduction.

Further, after the sound reproduction test is repeatedly repeated and improved in the sound synthesis output device, if the sound information is fixed as complete sound information, the sound information transmission path for reading the sound information is sealed. Thus, it is possible to prevent others from catching the digital voice information, and to maintain the confidentiality of the information.

If the digital voice information is stolen while the voice-synthesizing output device is equipped with the memory-integrated voice-synthesizing module and the voice-reproduction test is repeated and the improvement is repeated,
Alternatively, in a case where the sealing of the voice information transmission path after fixing the voice information is released by a chemical or the like and the digital voice information is stolen, the aforementioned data conversion means provided in a part of the voice information transmission path, Even if the voice information in the memory LSI is copied to another person, if the encryption rule is not decrypted, it becomes difficult to copy the voice synthesis module with good sound quality by copying, which also leads to protection of copyright.

A memory LSI is mounted on the front surface of the printed circuit board of the memory-integrated voice synthesis module, and a voice synthesis LSI is mounted on the back surface of the voice synthesis module so as to overlap each other.
The I address bus, data bus, and the like are vertically connected to each other via through holes in the printed circuit board. As a result, the mounting density of the printed circuit board is improved, and the memory-integrated speech synthesis module can be reduced in size and cost.

Further, the two LSIs are substantially the same size, and their arrangement directions are such that the pin arrangement of the address bus and the data bus of both LSIs are as close to each other as possible in the vertical direction with the printed circuit board interposed therebetween. When the direction in which the LSI is arranged is determined, the wiring pattern space can be further reduced, and the size can be further reduced.

[0027]

Embodiments of the present invention will be described below. (A) FIG. 1 is a structural diagram showing one embodiment of the memory-integrated speech synthesis module of the present invention. A rewritable memory LSI such as an EPROM is mounted as the storage element 3 on the front surface of the printed circuit board 1 of the voice synthesis module, and the voice synthesis LSI 2 is mounted as the voice synthesis element 2 on the back surface.
Implement In such a configuration, the arrangement directions of both LSIs are as follows.

The storage element 3 and the speech synthesis element 2
It is almost the same size depending on the type of package,
The arrangement direction of the two LSIs is determined so that the corresponding signal pins of the audio information address transmission path 5 and the audio information transmission path 4 of both LSIs are closest to each other in the vertical direction. At this time, a signal pattern can be easily connected in the vertical direction between the front surface and the back surface via the through hole of the printed circuit board 1, so that the pattern mounting space can be significantly reduced. For this reason, the pattern mounting density can be greatly increased as compared with the conventional speech synthesis module of FIG.

(B) The storage element 3 is a 16M BI
Large-capacity memory LSIs such as T and 64M BIT are being developed. With the progress of semiconductor manufacturing technology, the degree of integration has been improved year by year and the capacity has been increased, so that the storage amount of voice information can be physically increased and the market demand for higher sound quality and longer playback time can be sufficiently satisfied.

(C) The voice information is recorded by an announcer or the like in a recording studio, and the recorded voice message is once recorded by a tape recorder, and then analyzed and edited by various voice analysis methods using a voice analysis tool. For example, the digital audio information is written into the memory LSI by an EPROM writer or the like.

(D) Memory LS in which voice information is written
I is mounted on the surface of the printed circuit board 1, and the entire voice synthesis module including the voice information transmission path 4 for reading voice information is sealed with, for example, an epoxy-based resin in order to maintain circuit confidentiality and protect the copyright of voice information. It is molded with the stopper 6. Since the external signal pins 7 can be configured so that the audio information address transmission path 5 and the audio information transmission path 4 of the storage element 3 are not output to the outside for security reasons, the number of signal pins is significantly reduced, and handling is easy. become.

The memory-integrated speech synthesis module of the present invention is mounted on a printed circuit board in a speech synthesis output device for, for example, FA, OA, or service fields, and is used as a device for speech reproduction.

(E) FIG. 2 is a block diagram showing one embodiment of the memory-integrated speech synthesis module of the present invention. The speech information is stored in, for example, an EPROM as the storage element 3, and is used as the speech synthesis element 2, for example, in accordance with the sampling clock determined by the transmitter 8.
SI operates. The voice information address on the voice information address transmission path 5 is converted by the data conversion means 12 for performing address conversion according to a special encryption rule. The voice information corresponding to the converted voice information address is read to the voice information transmission path 4, converted by the data conversion means 12 for converting the voice information according to a special encryption rule, and read into the voice synthesis LSI 2. The converted voice information is synthesized by the voice synthesis LIS2, converted into an analog voice signal waveform, and output. This sound output signal is subjected to frequency characteristic conversion through a filter circuit 9 inside or outside the sound synthesis LSI 2, and then sound is output from, for example, a speaker 11 via an amplifier circuit 10 mounted on a printed circuit board in the sound synthesis output device, for example. You.

(F) The data conversion means 12 is a circuit for converting the information on the voice information address transmission path 5 or the voice information transmission path 4 into different information according to a special encryption rule. Are converted into various types of codes by bit array conversion or by combining a wafer / chip type programmable array logic (PAL), a mask ROM, and the like. The data conversion means 1 is added to protect the copyright of audio information and maintain confidentiality. Even if the sealing material 6 is removed and the contents of the storage element 3 are read and copied,
Since the second encryption (conversion) rule cannot be decrypted, unauthorized duplication as an audio reproduction device can be prevented.

On the other hand, when writing audio information into the storage element 3, the memory writing device writes digital audio information that has been subjected to inverse data conversion in advance. By this inverse conversion and preprocessing, the audio information read from the storage element 3 is returned to the original normal audio information by passing through the data conversion means 12, and is output with high sound quality.

(G) As another embodiment of preventing duplication, the present invention can easily implement frequency conversion characteristics of the filter circuit 9 as special conversion characteristics so that frequency analysis cannot be easily performed. For example, if the filter circuit 9 has a conversion characteristic of attenuating the frequency component of the frequency F or higher, the frequency component of the frequency of F or higher is preliminarily amplified at the time of the sound recording by the announcer in anticipation of the attenuation, and the sound is recorded. analysis,
By writing the edited audio information to the storage element 3, it becomes possible to implement an encryption (conversion) rule using frequency characteristics.

(H) FIG. 3 is a structural view showing another embodiment of the memory-integrated speech synthesis module of the present invention. FIG.
Similarly to the above, the storage device IC socket is mounted on the front surface of the printed circuit board 1 of the voice synthesis module instead of the storage device 3 instead of the storage device 3, and the voice synthesis device 2 is mounted on the back surface. According to the above configuration, the storage element 3 can be freely attached and detached via the IC socket 13, so that when the audio information is corrected or changed in the trial production stage of the audio information, the information in the storage element 3 can be easily rewritten.

Therefore, there is no waste of parts due to disposal or the like that occurs when information is rewritten as in the conventional example of FIG. As in FIG. 1, in order to maintain confidentiality of the data conversion means 12, the filter circuit 9, and the like, the I.
The whole is molded except for the C socket 13. The IC
Even if the storage element 3 is pulled out from the socket 13 and the audio information is copied, it is difficult to analyze the data conversion means 12 as in (f) above, so that duplication as an audio reproduction device can be prevented.

The configuration shown in FIG. 3 is used for a prototype experiment used in a transition period until the voice information is fixed, and the configuration shown in FIG. 1 is used for mass production after the voice information is finally fixed. The same printed circuit board can be shared. Since the memory-integrated speech synthesis module of the present invention can be used alone as a device for speech reproduction,
It is mounted and used on a printed circuit board in a speech synthesis output device for A, OA, service, game, or any other industrial field.

FIGS. 4 and 5 are sectional views showing another embodiment of the memory-integrated speech synthesis module of the present invention. When the storage element 3 is expanded in FIG. 1, it is possible to enlarge the printed circuit board 1 and mount a plurality of the storage elements 3 on the same substrate.

On the other hand, as shown in FIG.
6, the expansion storage element 14 is mounted, and the extension connector 15 is mounted.
By connecting a signal between the storage element 3 and the expansion storage element 14 via the CPU, the size of the speech synthesis module can be reduced as much as possible.

(D) As shown in FIG. 5, for example, a DIP type EPROM is used as the storage element 3, the signal pins of the EPROM 3 are mounted on the thin printed circuit board 1 from the front side, and voice synthesis is performed on the back side. According to the present invention, it is also possible to easily implement a molding with a sealing material 6 such as an epoxy resin after mounting a wafer-shaped speech synthesis LSI chip as the element 2.

In this case, since a part of the external pins of the speech synthesis module can be shared by the signal pins of the EPROM 3,
The audio information can be written by directly setting it in a commercially available EPROM writer, and the dedicated IC socket 13 shown in FIG. 3 becomes unnecessary.

(G) As another embodiment, a plurality of storage elements 3 in FIG. 3 are not mounted on the printed circuit board 1 but are mounted on a printed circuit board in an external voice synthesizing output device. A board connection connector with an external connection cable is mounted on the portion. The voice synthesizing element 2 of FIG. 3 and a plurality of storage elements on the external printed circuit board can be easily connected from the present invention through the board connecting connector with the external connection cable. With such a configuration, the expansion of the storage element can depend on the mounting space of the external printed circuit board, and the reproduction for a long time is facilitated.

(ヲ) In addition, the storage element 3 is an EP
In addition to the ROM, a memory LSI such as a one-time PROM, a mask ROM, a flash memory, an SRAM, a DRAM, or the like is used, and the speech synthesis element 2 is a speech synthesis LSI of various systems such as a waveform encoding system, an analysis synthesis system, and a rule synthesis system. In addition, the present invention includes the use of a voice recording / reproducing LSI of the same various types. The shape of the memory LSI and the voice synthesis LSI may be in the form of a wafer chip or DI.
P, SDIP, SOP, SOJ, TSOP, FP, MF
Use of various LSI packages such as P, QFP, PLCC, and PGA is also included in the present invention.

[0046]

According to the present invention, the following effects are produced. (A) Since the memory element and the voice synthesizing element are arranged in the vertical direction to minimize the wiring pattern space, the printed circuit board can be downsized by high-density mounting, and a low-cost and compact voice reproducing device can be realized.

(B) When a low-cost, compact memory-integrated speech synthesis module is mounted on a printed circuit board in various speech synthesis output devices, the mounting space can be reduced, and the cost and size of the entire device can be reduced by saving space. Will be possible.

(C) Since the memory unit is not built in the voice synthesizer but is independent and can use a large-capacity storage device, it can satisfy the market demand for high-quality sound and long-time reproduction, and furthermore, by changing the voice information. Since rewriting to the storage element can be easily performed via the detachable IC socket, economical loss such as disposal can be eliminated.

(D) By providing data conversion means in a part of the voice information transmission path for reading voice information, it is possible to prevent unauthorized copying of the high-quality voice synthesis module by copying voice information in the storage element. Copyright protection.

(E) Furthermore, by molding the memory-integrated speech synthesis module with a sealing material, including at least the speech information transmission path, copying of speech information in the storage element can be prevented, and the speech information can be more strictly controlled. Confidentiality can be measured.

(F) Since the memory element and the voice synthesizing element are integrated as a module and the voice information address transmission path and the voice information transmission path are not output to the outside, the number of external input / output signal pins can be significantly reduced, and handling is possible. Simplicity and space saving can be downsized.

[0052]

[Brief description of the drawings]

FIG. 1 is a structural diagram showing an embodiment of a memory-integrated speech synthesis module of the present invention.

FIG. 2 is a block diagram showing an embodiment of a memory-integrated speech synthesis module according to the present invention;

FIG. 3 is a structural diagram showing another embodiment of the memory-integrated speech synthesis module of the present invention.

FIG. 4 is a sectional view showing another embodiment of the memory-integrated speech synthesis module of the present invention.

FIG. 5 is a sectional view showing another embodiment of the memory-integrated speech synthesis module of the present invention.

FIG. 6 is a component layout diagram of a conventional speech synthesis module.

FIG. 7 is a diagram showing the internal configuration of a conventional speech synthesis LSI with memory.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Printed circuit board 2 Voice synthesis element 3 Storage element 4 Voice information transmission path 5 Voice information address transmission path 6 Sealing material 7 External signal pin 8 Oscillator 9 Filter circuit 10 Amplifier circuit 11 Speaker 12 Data conversion means 13 Storage element attaching / detaching means 14 Expansion Memory device 15 Extension connector 16 Extension printed circuit board 17 Wiring pattern mounting space 18 Voice synthesis LSI with memory 19 One chip area of LSI 20 Voice synthesis circuit unit 21 Memory unit 22 Memory write control unit

Continuation of the front page (56) References JP-A-59-98598 (JP, A) JP-A-61-239300 (JP, A) JP-A-64-1000 (JP, A) JP-A-3-76330 (JP) , A) (58) Field surveyed (Int. Cl. ⁷ , DB name) G10L 19/00 H05K 7/14

Claims (5)

(57) [Claims] A storage element (3) for storing audio information.
And a voice synthesizing element (2) for reading the voice information from the memory element and synthesizing and outputting the voice information, wherein the memory element (3) and the voice synthesizing element (2) are arranged so as to overlap each other, and have a laminated structure. And a voice information address transmission path for the storage element (3) and the voice synthesis element (2).
(5) A memory-type speech synthesis module, wherein signal pins corresponding to each other in a speech information transmission path (4) and the like are vertically connected and connected. 2. A memory integrated type according to claim 1, wherein the storage element (3) is arranged on one side of the printed circuit board (1), and the voice synthesis element (2) is arranged on the other side so as to overlap with each other in the vertical direction. Speech synthesis module. 3. The memory-integrated speech synthesis module according to claim 1, wherein at least a speech information transmission path (4) for reading speech information from the storage element (3) to the speech synthesis element (2) is sealed. . 4. A storage device (3) to a speech synthesis device (2).
A data conversion means (12) for converting voice information into another information is provided in a part of a voice information transmission path (4) for reading voice information, and the voice information is transmitted through the data conversion means (12). 4. A memory-integrated speech synthesis module according to claim 1, 2 or 3. 5. The memory-integrated speech synthesis module according to claim 2, wherein the storage element is disposed on one surface of the printed circuit board via the storage element attaching / detaching means. .