JPS602999A - Voice recorder - 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 (a) Industrial Application Field The present invention analyzes an input audio signal to convert it into a code string with information compression, and decodes this code string in the synthesis mode to generate audio. This invention relates to an audio recorder that synthesizes and outputs signals.

(b) Prior art Most conventional audio recorders were constructed using tape recorder devices that used magnetic tape to record analog signals, but in recent years various codes have been developed that can compress audio signals. A system has been developed that allows audio recorders to store audio in a memory in the form of a digitally encoded signal with a relatively small amount of information.

However, considering that various types of memory such as semiconductor memory are still expensive, it is desirable to save memory for storing encoded audio signals. There are quite a few silent periods, and for this reason, the entire silent period is encoded in the same way as the sound period, which has the disadvantage of wasting memory.

(C) Object of the Invention The present invention provides an audio recorder that can save memory for storing code strings of audio signals.

B) Structure of the Invention The audio recorder of the present invention is provided with silence detection means for detecting a silence state in audio based on a code string obtained by analyzing an audio signal, and the silence detection means When a long period of silence is detected, the code string during this silence is compressed into a specific length of time and stored in memory.

(e) Embodiment FIG. 1 (al, tb+ is an external view of an embodiment of the audio recorder of the present invention and a key layout diagram of its operation section, and FIG. 2 is a block diagram showing the functional configuration of the device of the present invention. The configuration of FIG. 2 will be explained below with reference to FIG. a filter/amplifier circuit for removing high-frequency components above the audio band of the signal; and an A/D converter Q3 for sampling the audio signal obtained from the filter/amplifying circuit az and converting it into a digital value. , and a parameter extractor αa that extracts a pitch parameter P and an amplifier parameter A from the Percoll coefficient, which is a code string obtained by compressing information based on the digital audio signal from the A/D converter (131). (2) is a randomly accessible parameter RAM, in which each parameter K obtained from the parameter extractor (141) of the speech analysis section (1).

PlA is stored. (31 is an audio synthesis unit, which includes three synthesizers c for synthesizing digital sample values of the audio signal based on P and A with each parameter read from the parameter RAM 121, and the synthesizer C3). A D-A converter (2) that converts a digital sample string of an audio signal into an analog audio signal;
A filter amplifier circuit (to) that blocks sampling noise of the audio signal from 2), and a filter amplifier circuit (to) that blocks sampling noise of the audio signal from
It consists of a speaker and a speaker that emit synthesized speech based on the audio signal from (to). (4) is the voice analysis section (1) above.
, parameter RAM +21, and speech synthesis section (3)
This is a control unit that controls each operation of the control unit (4), and the control operations of this control unit (4) are executed based on operation signals corresponding to each key from the operation unit (5) as shown in FIG. 1(b). . (6) is a silence input detector, which detects that the value of the amplifier parameter A1, that is, the parameter A indicating the energy of the voice, is zero among the parameters %P and A obtained from the voice analysis section (1). A silent input detection signal indicating a silent state in which no input voice is present at this time is sent to the control section (4). (7) is a silence output detector, and the above parameter RAM +21
It detects that the value of the amplifier parameter A read from is zero, and sends a silent output detection signal indicating a silent state in which no output sound is present to the upper self-control unit (4). . (8) is the above parameter RAM
+21 is an external tape recorder detachably connected in parallel via an interface (9), and the magnetic tape of the recorder (8) has a parameter RAM +2.
1. Similar parameters can be stored.

Next, the operation of the audio recorder of the present invention will be described, mainly focusing on the operation of the control section (4).

(1) Recording mode First, operate the index key ← of the operation unit (5) as shown in Fig. 1 (bl) as, for example, [3, d, and enter the index number for the audio to be recorded. Next, operate D Ishi Kensetsu of mode key 6, input voice to microphone αB, and when inputting this voice is finished, select mode key 62 mouth I [[What to operate. Therefore, control section (4) receives the 'RFC' signal and operates the voice analysis section (1) and the parameter RAM (21);
The speech analysis unit (1) derives the time series of P and A for each parameter based on the input speech, and converts this into 'S'I'OP
'The parameter RAM +21 will be set until a signal is obtained.

At this time, the control section (4) detects the silence input detector (6).
When a silent input detection signal indicating a silent state is received from , the time during which this signal is continuously generated, that is, the time when the sound is interrupted, is measured, and this measured value is set to a specific value, for example, 0.5.
When the second is reached, the writing operation of the parameter RAM (2) is stopped, and the writing operation is restarted when the silent input detection signal disappears. in particular,
The control unit (4) is provided with a counter function, and if the audio analysis unit (1) detects that the audio is in a steady state and obtains the amplifier parameter A every 20m8eQ, the silent input detector (6) determines that the audio is in a steady state. By incrementing the above counter function with the silent input detection signal (A=0) obtained every 20 mBeQ and cooling it with (A~0), when the count value reaches 25, that is, ,0
．． When (A = 0) continues for 5 seconds, the writing operation of parameter RAM (21) is stopped. Therefore, for example, 3 seconds of silence between sentences and clauses in the input audio. Even if there exists the parameter RA,
The pitch parameter P is set to 1/6 seconds to the amplifier parameter A=0 to M+21 and the associated Percoll coefficient.
Without storing 50 sets continuously, this can be done for 0.5 seconds.
The parameters are compressed and stored into 5 sets, and P%A can be omitted from the remaining 125 sets of useless parameters.

(11) Playback mode First, operate the index key . . . of the operation section (5) as shown in FIG. receives the ``PLAY'' signal and sets the parameter RAM +21 and the speech synthesis section (3).
1P and A are read out from the parameter RA 'M +21 as the parameter corresponding to index number 12' and transmitted to the speech synthesis section (3).
), the audio is synthesized. The synthesized speech obtained in this manner does not have unnecessary interruptions such as silence for a specific period of time, such as 0.5 seconds, and is easy to understand for the listener.

In addition, if there is a silence state of 0.5 seconds or more in such a synthesized voice playback mode, it means that the synthesized voice has finished speaking, and in order to detect this end, a silence output detector ( 7) is provided. That is, the control section (
4) receives a silent output detection signal indicating a silent state from the silent output detector (7), measures the time during which this signal is continuously generated, that is, the time when the sound stops, and calculates this measured value. If it exceeds a specific value, for example 0.5 seconds, which is the same as the specific measurement value for the above-mentioned silent input detection signal, ' 5T
The above parameter RA is set in the same way as when the OP' signal is received.
M (21. This will stop the operation of the speech synthesis section (3). Specifically, the measurement processing of this silence output detection signal is performed by measuring the silence input detection signal using the counter function provided in the control section (4). It is executed in the same way as processing.

Thus, in the time series of the audio amplifier parameters A of the parameter RAM +21, there is a case where (A=O) exists for a specific period of time, for example, 0.5 seconds or more.
So, the amplifier parameter A at the time the audio ends
The end of the audio can be determined by the presence of 26 or more consecutive A=O in the blank memory area of the parameter RAM (2) thereafter, and the audio recorder can be automatically stopped.

(lI) Pause mode In the above-mentioned playback mode, operate the mode key 6 of the operating section (5) as shown in FIG. 1+1)l. At this time, the controller/L [+41 is ' PAU
When the SE' signal is received, the reading operation of the parameter RAM +21 is temporarily stopped, and when the PAUSFJ# signal is received again, the reading operation is restarted.

Therefore, the synthesized speech synthesized by the speech synthesizer +31 can be temporarily interrupted at any time, and this is very effective in aiding the dictation and understanding of long speeches. Become something. This temporary pause mode is also effective in the recording mode, and audio recording can be performed in several parts.

H External input/output mode First, the external input/output switch (to) as shown in Figure 1(b)
When the recording mode described above is set with
, the audio parameters P, , A are transmitted to the tape recorder (8) via the interface (9), and conversely, when the above-mentioned playback mode is set, the operating state is transmitted from the tape recorder (8). Interface (9)
Transmission is input via . Therefore, for a long speech that exceeds the memory capacity of the parameter RAM + 21 configured in the semi-central memory, the time series of %P and A for that parameter is stored on a magnetic tape with a large memory capacity. A large number of short sentences requiring random access can be stored in the normal parameter RAM (21), and each can be played back as desired.

In the device according to the embodiment of the present invention described above, the control unit (4) determines the specific dark value time when measuring the generation time of the silent output detection signal in the playback mode. Similarly to that for the detection signal, for example 0.
Although it is set to 5 seconds, silence input detection processing is not performed in the recording mode, and there is a silence state of 0.5 seconds or more in the time series of %P and A for each parameter at parameter RAM +21, for example. This does not apply to gathering places. That is, the parameters LP collected from the outside (♂p tape recorder (8)) by another audio recorder that cannot perform the silent input detection process.

When synthesizing audio by introducing A, a configuration is adopted in which the control unit (4) can variably set the specific threshold time for the silence output detection signal to a value longer than the normal audio interruption time, for example, about 5 seconds. If so, it is possible to reliably reach the '5TOP state at the end of the synthesized speech without interrupting the synthesized speech.

(F) Effects of the Invention As is clear from the above description, the audio recorder of the present invention is provided with a silence detection means for detecting silence in audio based on a code string obtained by analyzing an audio signal. When the silence detection means detects a silent state for a long period of time longer than a specific time, the code string for this silent state is compressed into a specific time, so the memory stores the code string for the sound period as well as the compressed code string. The code string of the silent period that has been read is stored, and the code string of the useless long silent period can be omitted, making it possible to save a large amount of memory. Therefore, it greatly contributes to the miniaturization and cost reduction of this type of audio recorder.

[Brief explanation of the drawing]

FIG. 1 tal (bl is an external view of the audio recorder of the present invention,
1 and 2 are block diagrams showing the configuration of the audio recorder of the present invention. 16) Silence input detector, (7) Silence output detector. Figure 1 (b) ξ

Claims (1)

[Claims] (1) In the analysis mode, the input audio signal is analyzed and converted into a code string with information compression and stored in memory, and in the synthesis mode, the code string in the memory is combined to synthesize and output the audio signal. The audio recorder is provided with a silence detection means for detecting a silence state in the voice based on a code string obtained by analyzing the audio signal, and the silence detection means detects a silence state for a long period of time longer than a specific time. The voice recorder is characterized in that the code string in the canned sound state is compressed into a specific length of time and stored in the memory.