JPS6146920B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an analog signal reproducing device that stores analog signals such as audio in a memory and reproduces them, and particularly relates to an analog signal reproducing device that can faithfully reproduce a large amount of information by effectively utilizing the storage capacity of the memory. The present invention aims to provide a signal reproducing device.

Various speech synthesis devices have already been proposed that reproduce speech from information stored in semiconductor memory. Speech synthesis devices that have been put to practical use read out memory at a constant speed and synthesize speech using the read output information. For this reason, the storage elements of the memory are not always effectively utilized, and a large amount of information can be stored and reproduced.

In other words, as is well known, speech is divided into vowels and consonants. Vowels can be expressed as waves with a relatively low constant frequency, but consonants contain high frequency components at the beginning of the sound, and then become vowels. To clearly reproduce consonants, it is necessary to faithfully reproduce the onset of the sound. To faithfully reproduce high-frequency sounds, it is necessary to read a lot of information from memory. On the other hand, vowels do not require a large amount of information from memory because they are constant frequency components with relatively low frequencies. Therefore, since the reading speed of the memory is conventionally determined based on the reproduction of consonants, the memory is wasted for vowel parts. The functions required of a speech synthesizer are to memorize a lot of information,
It is desirable to be able to reproduce long words.

An object of the present invention is to provide an analog signal reproducing device that can store and reproduce as much information as possible.

In this invention, the frequency information of the audio is obtained from the audio information read from the memory, and the frequency information is used to change the reading speed of the memory.

Therefore, according to the present invention, a large memory storage area is provided for analog signals having high frequency components, and a small memory storage area is provided for analog signals having low frequency components. This makes it possible to effectively utilize the storage capacity of .

An example in which the present invention is applied to a speech synthesis device will be described below in detail with reference to the drawings.

FIG. 1 shows an apparatus for writing analog signals into memory. 101 indicates a memory. As the memory 101, for example, a semiconductor PROM or RAM can be used. 102 is a microphone, 103 is an amplifier, 104 is an AD converter, and 105 is a delay circuit. That is, the audio signal captured by the microphone 102 is amplified by the amplifier 103, and the audio signal is transmitted to the A-D converter 104.
Convert to digital code by memory 101
If it has a storage capacity of, for example, 8.times.8K bits, then the AD converter 104 is an AD converter that converts into a 6 to 7 bit digital code, for example.

On the other hand, 106 is a frequency information generator. This frequency information generator 106 includes means 107 for converting the frequency of the analog signal stored in the memory 101 into a DC signal, and A-- for converting the DC signal from analog to digital.
D converter 108. The means 107 for converting the frequency of an analog signal into a DC signal is provided with an AGC circuit 109 at the first stage, and this AGC circuit 109 converts the analog signal supplied from the amplifier 103 into a signal with a constant amplitude. The signal is provided to a frequency discriminator 110. For this frequency discriminator 110, for example, a high-pass filter having frequency characteristics as shown in FIG. 2 can be used. The output of the frequency discriminator 110 is applied to a rectifying and smoothing circuit 111 to obtain a DC signal corresponding to the frequency of the analog signal.

The DC signal output from the rectifying and smoothing circuit 111 is given to the AD converter 108 for AD conversion. In this example, a case will be explained in which frequency information is converted into a 2-bit digital code. The 2-bit digital code allows the frequency state of the input signal to be quantized into four states. Therefore, in the frequency characteristic of the frequency discriminator 110 shown in FIG. 2, for example, the A-D converter 108 converts the state of 5KHz or less into a code of "0,0", and the state of 5 to 8KHz into a code of "1,0". The state of 8 to 10 KHz is converted to a code of "0, 1", and the state of 10 KHz or more is converted to a code of "1, 1".

This AD conversion output is supplied to the memory 101 for storage, and is also supplied to the DA converter 112.
The digital code quantized into two states is converted back into an analog voltage, and the analog voltage is supplied to a voltage controlled oscillator (hereinafter referred to as VCO) 113 to generate an oscillation signal with a frequency corresponding to the frequency information stored in the memory 101. obtain. The frequency of this oscillation signal is, for example, 5KHz for "0,0", 2KHz for "1,0", 10KHz for "0,1", and 10KHz for "1,1".
It can be selected as 16KHz. This oscillation signal is applied to the A-D converters 104 and 108 and used as a timing signal for A-D conversion, and is also supplied to the address counter 114,
1 address signal is generated, and the A-D converter 10
Audio information output from 4 and A-D converter 108
Frequency information outputted from the VCO 113 is stored in an 8-bit storage area at each timing of one clock. In this example, since the frequency information is converted into a 2-bit digital code, the audio information is stored in the memory 101 as a 6-bit digital code.

According to such a writing device, the microphone 1
Depending on the frequency of audio information input from 02
The oscillation frequency of VCO113 is controlled, for example 10K
When a signal with a high frequency component of Hz or higher is input, the VCO 113 oscillates at a frequency of 16KHz, so the audio information of that high frequency component is sampled at a repetition rate of 16KHz, A-D converted, and stored in the memory 1.
01. Therefore, signals with high frequency components, such as the rising edge of a consonant, are A-D converted at high speed, and the A-D converted digital code is stored in the memory 101 in synchronization with the A-D conversion operation.
can be memorized. Also, when the end of a consonant changes to a vowel, the VCO changes depending on its frequency.
The oscillation frequency of 113 changes to a lower frequency, and as a result, the AD conversion operation becomes slower, and the writing speed to the memory 101 also becomes slower. Therefore, the amount of storage area used in the memory 101 can be reduced for low frequency signals.

When audio information is recorded in the memory 101 in this way, the memory 101 is transferred to the playback device 3 shown in FIG.
It is used by attaching it to the 01. This playback device 301 can be used, for example, for in-car broadcasting of a one-man bus.

Next, the reproducing apparatus shown in FIG. 3 will be explained.
An address counter 302 is connected to the address terminal of the memory 101.
A clock pulse whose frequency changes is applied from the VCO 303. 304 is a DA converter, which converts the 2-bit frequency information read from the memory 101 from analog to digital, and converts the converted output voltage into VCO3.
Controls the oscillation frequency of 03. The oscillation frequency of the VCO 303 is set to correspond to the oscillation frequency of the VCO 113 explained in FIG. In other words, when the frequency information is "0, 0", it oscillates at 5KHz, and "1,
8KHz when "0", 10KHz when "0,1",
Configure it so that it oscillates at 16KHz when it is "1, 1".

With this configuration, the memory 101 can read out the frequency information stored therein at a speed corresponding to the frequency information stored therein. The audio information read from the memory 101 is DA converted by a DA converter 305, smoothed by a filter 306, and then sent to a loudspeaker 307.
and outputs audio from the speaker 308.

This invention claims a reproducing apparatus shown in FIG. This playback device 301 is suitable for use as, for example, an in-vehicle guidance broadcasting device for a one-person bus or a destination guidance device for a bus. In other words, since bus in-board guidance or destination guidance must be prepared for each bus route, there are a large number of types of broadcast guidance even when looking only at one bus company. Although there are many types of content, the amount of memory required is small. In contrast, commercially available speech synthesizers require large-scale equipment to create a ROM for speech synthesis. Therefore, when creating a large number of ROMs for speech synthesis with one fixed content,
Although the cost per ROM can be made sufficiently low, in the case of a large variety of products and a small quantity, it is necessary to use a ROM for speech synthesis.
The manufacturing cost is very high.

Therefore, as in the present invention, the audio is simply A-D converted and stored in a semiconductor memory 10 such as PROM or RAM.
A sound generating device that stores data in the memory 101 and reproduces the sound by reading the stored data is suitable for manufacturing a wide variety of memories 101 in small quantities. However, according to the present invention, since the reading speed of the memory 101 is changed according to the frequency component of the audio, it is possible to store a large amount of audio information in a memory with a small capacity. Therefore, the audio information necessary for one route can be stored in one memory or a very small amount of semiconductor memory, about 2 or 3, so the memory 101 is made into a plug-in unit for the playback device 301. It can be structured to be detachable. If a ROM is used as the memory 101, a battery may be built into the unitized memory 101, and the memory may be retained by the built-in battery when removed from the apparatus.

By making the memory 101 detachable from the playback device 301 in this way, any vehicle can be used for any route by installing the memory 101 prepared for each route on the bus at the start of work.

In the embodiments shown in FIGS. 1 and 3, the frequency information is converted into a 2-bit digital signal and stored in the memory 1.
01, but as another method, for example, the frequency information of the input audio signal is converted into a 1-bit serial signal, and the serial signal is stored in the memory 10.
There is also a method of storing 1 bit per 1 bit. In this method, during reproduction, the serial codes read from the memory 101 may be sequentially loaded into a register, and the information loaded into the register may be used as parallel data.

Other embodiments of the present invention are shown in FIG. 4 and below. In the example of FIG. 4, the filter 306 has a variable filter configuration, and the variable filter 306 has a DA converter 3.
04 is supplied, and the frequency characteristics of the filter 304 are controlled to be suitable for the audio information read from the memory 101.

With this configuration, even if the frequency of the audio information read from the memory 101 changes, both high frequency components and low frequency components can be smoothed in an optimal state, and the clarity of the audio can be improved. .

The example shown in FIG. 5 shows a case in which the audio information is stored in the memory 101 in high-pitched and low-pitched tones, and the high-pitched and low-pitched tones are read out separately and reproduced by adding them together. In other words, 305A is a DA converter for high-pitched sounds, 3
05B is a bass DA converter. For example, 4 bits out of 8 bits can be used for high-pitched sounds, and 2 bits can be used for low-pitched sounds. The conversion outputs of these DA converters 305A and 305B for treble and bass are supplied to a treble filter 306A and a bass filter 306B.
The configuration is such that the outputs of A and 306B are added and supplied to the public address system 307.

In the case of such a configuration, even if a signal with a high frequency 602 is superimposed on a low frequency signal 601 as shown in FIG. 6, for example, the low frequency signal 601 and the high frequency signal 602 are It is sufficient to separate the signals 601 and 602 using a filter, and separately perform AD conversion on the separated signals 601 and 602 and store them in the memory 101. In this case as well, the memory 101 depends on the presence or absence of the high frequency signal 602.
It is assumed that changing the writing speed to is performed in the same manner as described above.

In this way, the treble and bass are separated and the memory 101
By storing the data in the 1000 and reading it out and reproducing it as sound, the bass signal 601 and the treble signal 602 can be faithfully reproduced as sound.

In other words, when the signal shown in FIG. 6 is A-D converted by just one A-D converter, the level of the high-pitched signal 602 is sufficiently lower than the level of the low-pitched signal 601, so the high-pitched signal 602 components will not be faithfully A-D converted. This results in the inconvenience that high-pitched sounds cannot be faithfully reproduced. Therefore, as explained in Figure 5, the bass and treble are separated and
By converting the signals into D, storing them in the memory 101, reading them separately, and superimposing them on each other, bass and treble sounds can be faithfully reproduced.

The example shown in FIG. 7 shows a case where the slope value, that is, the differential value of the audio signal is stored in the memory 101 after AD conversion, and the audio is reproduced using the readout signal. Therefore, when storing audio information in the memory 101, a differentiation circuit is provided before and after the AD converter, the audio signal is differentiated by the differentiation circuit, and the differentiated signal is converted from AD to the memory 10.
1. In this case, frequency information of the audio signal can be obtained from the magnitude of the differential value. All 8 bits are thus available for storing audio information. According to the 8-bit digital code, a total of 256 types of differential values can be converted from analog to digital. Therefore, for example, 256 levels of A
- The D transform can be divided into two, with 0 to 128 assigned to positive slope values and 128 to 256 assigned to negative slope values.

The digital value read from memory 101 is D-
After A conversion, the voltage is supplied to a variable lamp voltage generator 701. This variable lamp voltage generator 701 is D-
With a certain value of the analog voltage output from the A converter 305 as a reference, positive and negative voltages are
This circuit generates lamp voltages with 128 different slopes. The waveform of the signal output from the variable lamp voltage generation circuit 701 is a broken line voltage waveform as shown in FIG. This polygonal voltage signal is smoothed by a filter 306, and the same as described above is applied to a loudspeaker 307.
The sound can be reproduced by amplifying the signal and supplying it to the speaker 308.

In this case, the 8-bit digital information includes frequency information of the audio signal. In other words, when the positive and negative slope values are large, it can be determined that the audio frequency is high. Yotsute memory 10
The digital value read from 1 is supplied to a decoder 702, the decoder 702 determines whether the frequency of the audio information is high or low, and converts the determination result into, for example, a 2-bit digital code and outputs it.
The conversion output is converted to an analog voltage by the DA converter 304, and the VCO 30
3, it is possible to change the reading speed of the memory 101 according to the frequency of audio information.

As explained above, according to the present invention, the frequency of the audio information stored in the memory 101 is detected, and the reading speed of the memory 101 is changed in accordance with the detected frequency. The amount of memory elements used can be reduced. In addition, since the memory 101 is read out at a faster frequency only in the rising part of the voice, especially the consonant, there is little deterioration in the clarity of the voice, making it possible to reproduce voices that are easy to hear. A speech synthesizer can be obtained.

[Brief explanation of the drawing]

FIG. 1 is a block diagram for explaining the memory storage method used in the present invention, FIG. 2 is a graph for explaining the operation of FIG. 1, and FIG. 3 is an implementation of the analog playback device of the present invention. FIG. 4, FIG. 5, and FIG. 7 are block diagrams showing other embodiments of the present invention. FIG. 6 is a waveform diagram for explaining the operation of FIG. 5. FIG. 8 is a block diagram showing an example. 7 is a waveform diagram for explaining the operation of FIG. 7. FIG. 101...Memory, 304, 305...D-A converter, 306...Filter, 307...Sound amplification device, 30
3...VCO, 302...Address counter.

Claims (1)

[Claims] 1. In a device that reads digital data stored in a memory, performs D-A conversion, and reproduces an analog signal, frequency information is stored for each digital data in the memory, and frequency information is read from the memory. An analog signal reproducing device comprising means for changing the reading speed of the memory according to the frequency information.