JPH04161999A - Voice record reproducing device - Google Patents

Abstract

PURPOSE:To effectively utilize a voice memory by adding a silent section of a given period to voice data read out from a voice memory means based on silent information when the voice data are read out from the voice memory means. CONSTITUTION:First and second silence controllers 4-1 and 4-2, first and second selectors (SEL) 4-3 and 4-4, data correction device 4-5, and silence detector 4-6 are provided. A voice signal in a silent section is not memorized in a voice memory means 2-9, and only voice data in an effective voice section other than the silent section is stored in the voice memory means 2-9. Moreover a length of the silent section is detected with the silent section detection means 4-6, and information showing the length of the silent section is retained. The silent section is added to voice data read out from the voice memory means 2-9 when a voice is reproduced, consequently a voice phrase can be outputted at a timing difference same as at the time of recording. This can effectively utilize the voice memory 2-9.

Description

[Detailed Description of the Invention] [Purpose of the Invention (Industrial Application Field) This invention relates to a voice recording and reproducing device, and in particular to a voice recording and reproducing device used in a human voice response device that outputs predetermined voice data in response to input voice. The present invention relates to a recording/reproducing device.

(Prior Art) In general, a voice input response device provides a service of returning a predetermined voice message to the caller in response to voice input from the caller. is stored in advance in the audio memory.

A known method for recording and reproducing voices in this voice memory is to record and edit raw human voice information in advance, and then connect and output the voice information. The second method is called the recording and editing method, and is superior in the naturalness of pronunciation.

Conventionally, an audio recording/reproducing apparatus using this method has been constructed as shown in FIG.

When the audio recording and reproducing device of FIG. 3 records and reproduces the audio signal of FIG. 4, it first responds to the switching signal to record/reproduce
The playback switching circuit 2-10 supplies a control signal to the CPU 2-7 to set the audio recording/playback device to a recording state.

Then, recording starts at timing ti in Figure 4, and at timing t2, for example, "Tomorrow's weather is..."
The first audio phrase begins to be input.

A period a has already elapsed from timing tl to timing t2. The first audio phrase continues for a period of time, and then when a period of time C has elapsed, for example, "・
The second voice phrase "It's going to be sunny" begins to be input at timing t4, and ends after the period d has elapsed.

During the period from timing tI to t5, an analog audio signal is always input to the analog/digital (A/D) converter 2-1, and the analog audio signal is input to the serial/digital (A/D) converter 2-1.
Parallel (S/P) converter 2-2 and register (R
EG) 2-3, it is stored as parallel digital audio data in the audio memory 2-9. In this case, all the audio signals during the period from timing t1 to t5 are considered to be valid data, so the audio memory 2-9
In this case, as shown in FIG. 5, both silent data and actual voice phrases are stored.

On the other hand, during playback, the contents of the audio memory 2-9 are sequentially read out, and the contents of the register (REG) 2-6 and parallel/
Serial (P/S) variable m W 2-5 and digital/
The signal is converted into an analog audio signal via an analog (D/A) converter 2-4, and is reproduced by an amplifier and a speaker as necessary.

2, in the conventional audio recording and reproducing apparatus, the audio signals in silent sections a and C are stored in the audio memory 2-9 together with the audio phrases in sections S and d.

Is the silent section important in the sense that it makes a difference in the output timing between audio phrases during playback?Since the audio signal in the silent section is silent even during playback, it is originally stored in the audio memory 2-9. It is a meaningless signal.

For example, the total silent interval (a + c) is several tens to hundreds [ms
When it comes to J, even if sampling is performed at 8 bits and 8 KHz, a considerable portion of the audio memory 2-9 will be occupied by silent sections. In particular, in the case of an audio signal containing a large number of phrases, a silent interval is inserted between each phrase, so the amount of silent interval stored in the audio memory 2-9 is also increased, and many of the audio memories 2-9 are Memory space will be wasted.

Also, to reduce this waste of memory space,
There is also a method of editing the contents of the audio memory 2-9 to reduce silent sections, but in this case, a lot of time is spent on editing.

(Issue 8 to be Solved by the Invention) Conventionally, since audio signals of silent sections are also stored in the audio memory, there has been a drawback that a large amount of memory space in the audio memory is wasted.

This invention was made in view of these points,
It is an object of the present invention to provide an audio recording and reproducing device that can realize effective use of audio memory by making it possible to output different audio phrases during playback without storing audio signals in silent sections in audio memory. do.

[Structure of the Invention] (Means and operations for solving the three problems) The audio recording and reproducing device according to the present invention detects a silent section of human voice,
a silent interval detection means for generating silence information indicating the occurrence of a silent interval and the length of the silent interval; a means for retaining the silence information generated by the silent interval detecting means; voice storage means for storing only voice data; and means for adding a silent section of a predetermined period to the voice data read from the voice storage means based on the silence information when reading the voice data from the voice storage means. It is characterized by comprising the following.

In this audio recording and reproducing apparatus, the audio signal within the silent section is not stored in the audio storage means, and only the audio data in the valid audio section other than the silent section is stored in the audio storage means. Moreover, the length of the silent section is detected by the silent section detecting means, and information indicating the length of the silent section is held. When reproducing the audio, this silent section is added to the audio data read from the audio storage means, so the audio phrase can be output with the same timing difference as when recording. Therefore, it is possible to make outputs of audio phrases different during playback without storing the audio signal of the silent section in the audio memory, and it is possible to effectively utilize the storage capacity of the audio storage means.

(Example) Hereinafter, the present invention will be described in detail with reference to the drawings.

FIG. 1 shows an audio recording and reproducing apparatus according to an embodiment of the present invention. This audio recording and reproducing device is used in a voice input response device, and in addition to the conventional configuration shown in FIG.
NT, 1. Silence C0NT, 2) 4-1.4-2, 1st
and second selector (SEL) 4-3, 4-
4, data correction device 4-5, and silence detector 4-6
It is equipped with

The first silence controller 4-1 generates a silence flag indicating a silence interval and silence interval count information indicating the length of the silence interval, and the operation of generating the silence flag and silence interval count information is performed by the CP. U 2-7 and its output operation is controlled by Direct Memory Access Controller (DMAC) 2-8.

The second silence controller 4-2 detects the silence flag and silence section count information stored in the audio memory 2-9, and based on the detection result, performs communication between the CPU 2-7 and the direct memory access controller 2-8. control data transfer.

The first selector 4-3 is a first silent controller 4-3.
This is for selecting one of the silence flag and silence count information output from register 1 and the audio data from register 2-3 and supplying it to audio memory 2-9.
The selection operation is controlled by C, P U 2-7.

The second selector 4-4 selects either the output from the audio memory 2-9 or the output from the data correction device 4-5 and outputs it to the register 2-6, and controls the selection operation. is controlled by CPU 2-7.

The data correction device 4-5 stores in advance data to be sent during the silent period, that is, silent data corresponding to the silent state (for example, data in which all bits are "O'") and other noise data. Output data.

The silence detector 4-6 detects a silent section from an analog audio signal supplied from the outside and notifies the CPU 2-7. It is detected based on whether a state lower than the value voltage continues for a certain period of time or more.

Next, the operation of the audio recording/reproducing apparatus shown in FIG. 1 will be explained by taking as an example the case where an audio signal as shown in FIG. 4 is manually input.

This audio signal is sent to an analog/digital converter 2-1, a serial/parallel (S/P) converter 2-2, a register (
REG) 2-3, the signal is converted into digital data and supplied to the selector 4-3, as well as to the silence detector 4-6.

When the silence detector 4-6 detects that the section a in FIG. 4 is silent, it notifies the CPU 2-7 of this fact. At this time, the first silent controller 4-1 also has C.
It is notified from P U 2-7 that a silent section has occurred.

The CPU 2-7 controls the selector 4-3 to select the output from the silent controller 4-1. At the same time, the silence controller 4-1 generates a silence flag and starts counting the length of the silence section at a time point near timing t1 in FIG. 4.

This counting operation is performed by the silence detector 4-6 or
The process is executed until a voice phrase is detected, that is, until timing t2.

Furthermore, at a time point near timing t2, the CPU
2-7, the silence controller 4-1 is further notified by the silence detector 4-6 that the silence period has ended.

The silence flag and silent section count information generated by the silence controller 4-1 are stored in the second audio memory 2-9 by the joint operation of the CPU 2-7, the direct memory access controller 2-8, and the selector 4-3. As shown in the figure, the information is stored as a silence flag (a) and silence count information (b).

After the completion of the second operation, the first audio phrase in the section of FIG.
) 2-3 and selector 4-3, the data is stored in the audio memory 2-9 as digital audio phrase data (b) as shown in FIG.

After 2, when a silent state occurs at timing t3 in FIG. The silence flag (C) indicating the occurrence and the silence count information (C) indicating the length of the silence section C are stored in the audio memory 2.
−9, and further, near the timing [4,
The second audio phrase is digital audio phrase data (b
) starts to be stored in the audio memory 2-9.

Next, the operation when reproducing audio will be explained.

In case 2, the CPU 2-7 directly instructs the memory access controller 2-8 to sequentially read data from the audio memory 2-9.

First, a silence flag (a) and silence count information (a) are read out from the audio memory 2-9. When the second silence controller 4-2 detects the read silence flag (a), it notifies the CPU 2-7 of the existence of the silence flag (a). Furthermore, the second silence controller 4-2 also recognizes the value of silence count information (a).

When the CPU 2-7 is notified of the presence of the silence flag <a), it controls the selector 4-4 to select the output of the data correction device 4-5. In addition, the silent controller 4
-2 starts counting the value of the silence count information (a) again, and during the counting period, the silence data from the data correction device 4-5 is sent to the selector 4-4, the register 2-6, the parallel/serial CP/S) converter 2-5
, and output as a silent signal via the digital/analog converter 2-4.

2, when silence data for the value of the silence count information (a), that is, for the period a, is output, the silence controller 4-2 notifies the CPU 2-7 that the silence period a has ended. Notice. Upon receiving this notification, CPU2-7
controls the selector 4-4 so that the audio phrase (b) in the audio memory 2-9 is selected. Thereby, the audio phrase (b) is transferred to the register 2-6, the parallel/serial converter 2-5, and the digital/analog converter 2.
-4, it is output as an analog audio signal.

Thereafter, the silence controller 4-2 controls the silence flag (
C) and the silence count information (C) are detected again, similar to the above-mentioned silent interval a, silence data for the silent interval C is output, and then the second audio phrase (d) is output as an analog audio signal. is output as

As described above, in this embodiment, the voice signal within the silent section is not stored in the voice memory 2-9, but the voice phrases in the valid voice section other than the silent section are stored in the voice memory 2-9. Ru. Furthermore, the length of the silent section is indicated by the silent count information. When reproducing audio, this silent section is added to the audio data read from the audio memory 2-9, so it is possible to output an audio signal with the same waveform as the input audio waveform.

[Effects of the Invention] As described above, according to the present invention, it is possible to make a difference in the output of audio phrases during playback without storing the audio signal of the silent section in the audio memory. Effective use can be achieved.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an audio recording and reproducing apparatus according to an embodiment of the present invention, and FIG. 2 shows data contents stored in an audio memory provided in the audio recording and reproducing apparatus shown in FIG. 1. 3 is a block diagram showing the configuration of a conventional audio recording and reproducing device, FIG. 4 is a waveform diagram showing an example of an audio signal input to the audio recording and reproducing device of FIG. 3, and FIG. FIG. 2 is a diagram showing data contents stored in an audio memory provided in the audio recording/reproducing device shown in the figure. 2-9... Audio memory, 4-1, 4-2... Silent controller, 4-5... Data correction device, 4-6...
... Silence detector. Applicant's agent Patent attorney Takehiko Suzue Figure 4 Figure 5

Claims (1)

[Scope of Claims] A silent section detecting means for detecting a silent section of input audio and generating silence information indicating the occurrence of the silent section and the length of the silent section; Silence information generated by the silent section detecting means. a voice storage means for storing only voice data of a valid voice section of the input voice; and a voice storage means for storing a voice data of a predetermined period of silence based on the silence information when reading voice data from the voice memory means. An audio recording and reproducing apparatus comprising: means for adding to audio data read from the audio storage means.