JPS61177695A - Voice memory device - Google Patents

Abstract

PURPOSE:To reduce the amount of stored information to reduce the memory capacity by effecting PCM encoding under the state in which voiceless portions between the voice signals of predetermined units are compressed. CONSTITUTION:Syllables (b), (d) are converted by different PCM codes according to positive and negative voice ranges of the voice waveform. A code FF is allotted to a voiceless portion (a) before the syllable band to a voiceless portion (c) between the syllables (b) and (d) and the voice waveform added by a code representing the time duration of the portions of the (a) and (c) is coded by PCM. The time duration of the voiceless portions (a) and (c) are compressed regardless of the duration so that the information stored in the ROM is reduce owing to compression to two bytes. This results in a reduced capacity of the voice memory unit and reduced size of the voice memory device.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to an improvement in a voice memory device that stores voice waveforms in PCM code.

(Technical Background of the Invention) For example, a voice response device that uses a telephone line to provide various automatic notifications such as time signals, weather forecasts, reservation status, etc. is generally equipped with a voice memory device, and the voice memory device stores data in advance. The device is configured to selectively read the stored audio waveform information and sequentially reproduce it into audio, thereby transmitting a desired audio message.

Incidentally, in recent years, as a voice memory device used in a voice response device, there is a device in which a voice waveform is PCM encoded and stored in a memory. This type of voice memory device stores and reproduces voice waveforms as they are, so it is faster to reproduce than, for example, a device that stores the rules for voice synthesis and synthesizes voice based on these rules. It is characterized by extremely excellent sound quality.

[Problems with background technology]

However, conventional audio memory devices of this type store all audio waveforms as they are without any compression processing.
The CM is encoded and stored. This has the disadvantage that the storage capacity of the audio memory increases, leading to an increase in the size and cost of the device.

[Purpose of the invention]

The present invention makes it possible to encode silent parts between predetermined units of audio signals in a compressed state, thereby reducing the amount of stored information and reducing the storage capacity of the audio memory, thereby reducing the size and cost of the configuration. It is an object of the present invention to provide a voice memory device capable of achieving the following.

[Summary of the Invention] In order to achieve the above object, the present invention provides that the current PCM encoding method, which converts the positive audio signal region and the negative audio signal region of the audio waveform into different code strings, Focusing on the fact that the silence level of is expressed by the first and second codes, the audio waveform is assigned the first code representing the silence level to the silence part existing in a predetermined unit of audio signal, and each The silent portions existing between the predetermined units of audio signals are distinguished by assigning a second code, and the second code is assigned the time length of the silent portions existing between the predetermined units of audio signals. The second code is PCM encoded by adding a length code representing , and stored in the audio memory, and the second code is detected from the PCM code string read from the audio memory, and the second code is detected. When the length code is added to this code, a silent portion of the time specified by the length code is reproduced.

[Embodiments of the invention]

One of the currently used PCM encoding methods for audio waveforms is the μm LAW method. This method has non-linear analog-to-digital conversion characteristics, as shown in FIG. The OV (silence level) of the analog input is converted into two codes: the first code "7F" and the second code "7F". 2 and is represented by the symbol "FF".

Incidentally, FIG. 5(b) is an enlarged view of the conversion characteristic near OV.

This embodiment focuses on the fact that two types of codes are assigned to Ov of the analog input signal, and the first code is assigned to the silent part existing in a predetermined unit of the audio signal of the audio waveform, for example, a syllable. "7F" is assigned, and a second code "FF" is assigned to the silent part between each syllable, and a length code representing the time length of the silent part between the syllables is added to this second code. By doing so, the audio waveform is PCM encoded, and this PCM code string is read and read as an audio memory.
It is stored in only memory (ROM). for example,
When a speech waveform as shown in Fig. 3 is PCM-encoded and stored in the speech memory, the speech waveform at the beginning of the syllable and the second part is PCM-encoded as is. The silent parts A and A are encoded by assigning the code FF and adding a code representing the time length to the silent part A to this code F F .Therefore, each silent part A, The encoding result of
No matter how long the silent portion is, it will be 2 bytes, and the amount of information will be reduced accordingly.

FIG. 1 shows an example of the configuration of a voice memory device using a ROM in which voice waveforms are stored in this way. On the other hand, there is a gate circuit 3 consisting of AND gates 3a, 3b and an OR gate 3C that supply count-up pulses, and a parallel-serial (P/S) 4 that converts the signal form of the PCM code read from the ROMI from parallel to serial. , a detection circuit 7 consisting of AND gates 7a and 7b, which detects whether or not the PCM code read from the ROM 1 is "FF" indicating silence, and a silence reproduction circuit 11; The signal generation circuit 12 generates each control pulse for controlling the circuit section.

The silent reproduction circuit 11 is composed of a counter 8, flip-flops 5 and 6, an AND gate 9, and a three-state OR gate 10 gated by a control signal GS output from a signal generation circuit 12. R.O.
The time specified by the length code of the silent portion read from MI is set, this time is held in flip-flop 5.6, and a silent signal (OV) is output from three-state OR gate 1o during this period. .

Next, the operation of the apparatus configured as above will be explained. At this time, it is assumed that the ROMI stores, for example, a PCM encoded voice code as shown in FIG. still,
In the figure, 'FF' represents the silent part between syllables, 'AF', 'EF', 'AO', '0'.
9” is a code representing the time length of the silent portion, “
xx” is a phonetic code.

In this state, for example, when a response command signal CC arrives from the control circuit (not shown) of the telephone exchange, the signal generating circuit 12 first outputs a step pulse As as shown in FIG. becomes °゛00, and as a result, the code “FF” starts from address “OO” in ROM1.
is read out.

Subsequently, when the load pulse BS is output from the signal generation circuit 12, the code "FF" read from the ROMI is
is loaded into the P/S converter 4.

Further, the above code is guided to the detection circuit 7 and the AND gate 7a
"Fj" is detected, and as a result, a detection signal (") (" level) is output from the AND gate 7b, and the seven lip-flops 5 and 6 of the silence reproduction circuit 11 are set, respectively. When these 7 lip-flops 5 and 6 are set, the AND gate 3b of the gate circuit 3 is opened and the count-up pulse C8 generated from the signal generation circuit 12 is supplied to the address counter 2. The address "01" is accessed and the code "AF" representing the time length of the silent portion is read out. When the reading of the code "AF" is completed, the signal generating circuit 12 outputs a load pulse O8 and supplies it to the counter 8, whereby the code "AF" read from the ROM1 is input to the counter 8. be introduced. As a result, the counter 8 starts counting the clock pulses FS generated from the signal generating circuit 12, and when the count value reaches the value indicated by the symbol "AF", it generates a coincidence signal and flips the flip 7a. Reset knob 6.

Therefore, the flip 7O block 6 maintains the set state while the counter 8 is counting, and as a result, an "H" level signal is continuously output to the three-state OR gate 10 during the counting period. be done. Therefore, the three-state OR gate 10 sends out a silent code string having the time length specified by the code "AF". Note that during the transmission period of this silent code string, the flip-flops 5 and 6 are in the reset state and set state, respectively, so that each AND gate 3a of the gate circuit 3
, 3b remain closed, and as a result, the count-up pulse C8 is not supplied to the address counter 2, and the ROM
No new phonetic code is read from 1. That is, during the transmission period of the silent code, the readout of the voice code from the ROM 1 is in a standby state.

Then the count value of the counter 8 becomes 'A F'
When the 7-lip flop 6 is reset, the transmission of the silent code ends at this point, and the AND gate 3a of the gate circuit 3 becomes open. Therefore, the supply of the count-up pulse C8 to the address counter 2 is resumed thereafter, and as a result, the voice code "x x" stored at the next address "02" is read out from the ROM 1. This audio code "xx" is introduced into the P/S converter 4 in synchronization with the load pulse BS from the signal generating circuit 12,
After being converted from a parallel signal to a serial signal in synchronization with the shift pulse ES, it is outputted to a decoding circuit (not shown) via a three-state OR gate 10 as a PCMrI% sequence H8.

Similarly, from ROM1, the code ""F" representing the silent part is
The silence reproduction circuit 11 operates every time ``F'' is read out, and this silence reproduction circuit 11 reproduces the above code ``'' F F
'', a silent code string having a time length specified by the code representing the time length of the silent portion read out from the ROM 1 is created and output.

In this example, one of the two codes representing Ov is assigned to the silent part between syllables,
Moreover, since the audio waveform is PCM encoded by adding a code representing the time length of the silent portion after this code, the information after encoding is greater than that of conventional devices that encode the entire audio waveform as it is. This allows the capacity of the ROM 1 to be reduced, making the configuration simple and compact, and making the device inexpensive. In addition, by separating the code assigned to the silent part between syllables and the code assigned to the silent part existing within a syllable, it is possible to compress only the silent part existing between syllables, which improves efficiency. Compression can be performed. That is, the silent portions that exist within a syllable generally occur when the signal changes from positive to negative or from negative to positive, and therefore have a very short time length. For this reason, if such silent parts within a syllable are encoded in the same manner as the silent parts between syllables, the amount of information may increase, which is not efficient. On the other hand, since silent parts between syllables are generally long, it is extremely effective to reduce the amount of information in these silent parts. Furthermore, noting that the conventional μmLAW system expresses the ov of the analog signal with two types of codes, we assigned these two existing codes to each silent part, so we set a completely new code. As a result, there is an advantage that it can be realized extremely easily without changing the PCM code itself.

Note that the present invention is not limited to the above embodiments. For example, in the above embodiment, of the two codes representing the OV of the analog input, "'FF" is assigned to the silent portion between syllables, but "7F" may be assigned. In addition, the means for detecting the code assigned to the silent part between syllables, the means for reproducing the silent part, the application of the voice memory device, etc. can be modified in various ways without departing from the gist of the present invention.

〔Effect of the invention〕

As described in detail above, according to the present invention, it is possible to compress and encode silent parts between predetermined units of audio signals, thereby reducing the amount of stored information, thereby reducing the storage capacity of the audio memory. Accordingly, it is possible to provide an audio memory device that can reduce the noise, have a smaller structure, and lower the cost.

[Brief explanation of drawings]

1 to 4 are for explaining a voice memory device according to an embodiment of the present invention, FIG. 1 is a circuit diagram of the device, and FIGS. 2 and 3 are used to explain the operation. A timing diagram and a signal waveform diagram, FIG. 4 is a schematic diagram showing an example of a state in which encoded audio signals are stored in ROM, and FIG.
), (b) is μ-LA as one of the PCM encoding systems.
It is a figure which shows the conversion characteristic of W method. 1...ROM12...Address counter, 3...
Gate circuit, 4...Parallel-serial (P/S) converter, 5, 6
...Flip 70 knob, 7...detection circuit, 8...
Counter for silent playback, 9...And gate, 10.
... Three-state OR gate, 11... Silence playback circuit, 12... Signal generation circuit. Applicant's agent Patent attorney Takehiko Suzue [3I! P! l Figure 4 Figure 5 (a) Analog input (V) (b)

Claims (2)

[Claims] (1) Converting the audio waveform into different PCM code strings for the positive audio signal region and the negative audio signal region, and expressing the silence level of the audio signal using the first and second PCM codes.
In an audio memory device that encodes and stores audio waveforms using a CM encoding method, a first code representing the silence level is assigned to a silent portion existing in a predetermined unit of audio signal, and a first code representing the silence level is assigned to each predetermined unit of audio. A voice memory in which a voice waveform is PCM encoded and stored by assigning a second code to a silent portion existing between signals and adding a length code representing the time length of the silent portion to the second code. and means for detecting the second code from the PCM code string read from the audio memory;
an audio memory device comprising means for reproducing a silent portion of a time length specified by a length code added to the second code when the second code is detected. (2) The audio memory device according to claim (1), wherein the predetermined unit of audio signal is a syllable-based audio signal.