JP3252591B2 - Voice storage device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice storage device for storing digitally encoded input voice data for use in solid-state recording of an answering machine or the like.

[0002]

2. Description of the Related Art In recent years, in an answering machine, etc., a call received when an answering machine is received is recorded by using a solid-state element instead of recording on a conventional magnetic tape. In the recording of the call voice in the solid-state device, the voice signal is digitally encoded, and silence of the digitally encoded voice signal is detected, and voice data excluding the silent portion is recorded. As a result, more audio data can be recorded without requiring a large recording capacity of the recording medium.

FIG. 3 is a block diagram showing a conventional voice storage device. In FIG. 3, this example corresponds to an input audio signal Sa.
, A sound detection unit 1 that performs sound / silence judgment by detecting a soundless portion, a sound recording unit 2 that records the sound signal Sb after the sound / silence judgment, and a sound signal Sc recorded by the sound recording unit 2.
And an audio reproducing unit 3 for transmitting an output audio signal Sd from the audio reproducing apparatus. Further, there is provided a control unit 4 for transmitting a voice recording control signal Ca and a voice reproduction control signal Cb to the voice recording unit 2 and the voice reproducing unit 3 based on the voiced / non-voice determination information Se from the voiceless detecting unit 1.

Next, the operation of this conventional example will be described. First, in the case of recording, the silence detector 1 calculates the power of each frame of the input audio signal Sa. In this power calculation, when the power is high, it is determined that there is sound, and when the power is low, it is determined that there is no sound. Then, the sound signal Sb to which the sound / no-sound determination is added is output from the soundless detection unit 1 to the sound recording unit 2. At the same time, the sound / silence determination information S of the frame
e is output to the control unit 4. The control unit 4 counts the number of silence frames, and outputs to the audio recording unit 2 an audio recording control signal Ca for recording the number of silence frames and audio data of the sound determination frame. The audio recording unit 2 records the number of silent frames and the sound determination frame of the audio signal Sb by the input audio recording control signal Ca. In the case of reproduction, the control unit 4
Outputs the audio reproduction control signal Cb to the audio reproduction unit 3, and the audio reproduction unit 3 outputs silence data for the frame time from the number of silence frames of the audio signal Sc which is the recording data from the audio recording unit 2, The frame is played as is.

[0005]

However, in the conventional voice storage device, the silent voice detection unit 1 uses the input voice signal Sa.
Speech and silence determination performed by detecting a silent part from is only the power information of the frame, and the head of the fricative sound such as the syllabary of the Japanese syllabary is determined to be silent, and the beginning of the talk occurs,
There was a problem that high-quality voice storage was not possible.

The present invention solves such a conventional problem. When the speech power at the beginning of a speech, such as the Japanese syllabary, is monotonically increasing, the speech power is small, and it is determined that there is no sound. It is an object of the present invention to provide an excellent voice storage device that can prevent the beginning of a speech by a frame and store high-quality voice.

[0007]

According to a first aspect of the present invention, there is provided an audio storage device for detecting the power of digitally encoded audio data to determine whether the audio data is sound or silence. a sound determination unit, wherein the frame storage means for temporarily storing the most recent positive integer number of frames of digitally encoded audio data, said frame memory means positive stored in integer number of frames Power increase / decrease detection means for detecting and storing an increase / decrease in power between each adjacent frame, and the power increase / decrease detection when a frame changes from a silent frame to a sound frame based on the determination by the voiced / silent determination means. Positive integer stored in the means
Based on the number of frames power increase or decrease information, the voice frame only frames continuously power increases monotonically to a sound frame previous of said frame storage means positive integer number that are stored in the frame And voice recording means for recording in the voice storage means.

According to a second aspect of the present invention, in the voice storage apparatus, the voice / silence determining means determines that the voice data is a voice frame when the power exceeds the voice / silence determination value, and determines the voice data if the power does not exceed the voice / silence frame. And when the frame storage means changes from a silence frame to a speech frame, the power monotonically increases continuously before the speech frame in the positive integer number of frames in the silence frames before the speech frame. The monotonically increasing frame and the sound frame after the sound frame are recorded as audio data.

[0009]

The voice storage device according to the first and second aspects of the present invention temporarily stores the latest input voice only for a positive integer number of frames, and stores each of the stored positive integer numbers of frames. Changes in adjacent power are detected and stored. When the voice activity determination result is changed from a silence, based on the positive integer number of frames power fluctuation information stored, a positive integer number stored
It records the audio data to a sound frame previous of the frames to the frame only where power is increased monotonically continuously, in addition to the beginning of a positive integral number of voice frames. Therefore, when the voice power at the beginning of the speech is monotonically increasing, such as the syllabary of the Japanese syllabary, the speech power is small, and the beginning of the speech due to the frame determined to be silent is prevented.

[0010]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of a voice storage device according to the present invention.

In the following text and figures, the same components as those shown in FIG. 3 are denoted by the same reference numerals.

FIG. 1 is a block diagram showing the configuration of an embodiment of a voice storage device according to the present invention. In FIG. 1, this example shows a silence detecting section 1 for detecting a silence portion from an input audio signal Sa and performing a sound / silence judgment, an audio recording section 2 for recording an audio signal Sb after the sound / silence judgment, and The audio recording unit 2 and the audio reproduction unit 3 respectively output sound based on the audio reproduction unit 3 that sends out the output audio signal Sd from the audio signal Sc recorded by the recording unit 2 and the sound / non-speech determination information Se from the silence detection unit 1. A control unit 4 for transmitting a recording control signal Ca and an audio reproduction control signal Cb is provided. Further, the input audio signal Sa from the silence detection unit 1 is input, and the frame storage unit 5 that sends out the temporarily stored frame signal Sf, and outputs power increase / decrease information Sg from the frame signal Sf to the control unit 4. A power increase / decrease detection unit 6 is provided.

Next, the operation of this embodiment will be described. First, in the case of recording, the power of each frame of the input audio signal Sa is calculated by the silence detection unit 1, and if the power is large in this calculation, it is determined that there is sound. If the power is small, it is determined that there is no sound. At the same time, the input audio signal Sa
Are always stored in the frame storage unit 5 in sequence only with the latest positive integer number of frames (denoted by α as necessary). The power increase / decrease detection unit 6 detects and stores the power difference between adjacent frames of a positive integer number of frames from the frame signal Sf stored in the frame storage unit 5, and
The power increase / decrease information Sg is output to the control unit 4. The sound signal Sb to which the sound / no-sound judgment is added is output from the soundless detection unit 1 to the sound recording unit 2, and the sound / no-sound judgment information Se of the frame is output to the control unit 4. The control unit 4 determines the presence / absence of sound / non-sound
If e is a continuous silence frame, the number of silence frames continues to be counted, and when the speech / silence determination information Se changes from silence to speech, it is stored in the frame storage unit 5 from the power increase / decrease information Sg. In order to add a monotonically increasing frame whose power is monotonically increasing before a voiced frame in a positive integer number of frames to be added to the beginning of a voiced frame, "number of silence frames minus (minus) monotonically increasing Number of frames "and an audio recording control signal C for recording audio data of a sound frame to which a monotonically increasing frame is added.
a to the audio recording unit 2. The audio recording unit 2 receives the audio recording control signal Ca and generates “the number of silent frames−the number of monotonically increasing frames” and the monotonically increasing frame in the audio signal stored in the frame storage unit 5 from the frame signal Sf. Then, the voice data of the sound determination frame is recorded in addition to the head of the frame determined to be sound frame.
In the case of reproduction, the control unit 4 outputs a sound reproduction control signal Cb to the sound reproduction unit 3, and the sound reproduction unit 3 calculates the number of silence frames recorded in the sound recording unit 2 from the sound signal Sc recorded by the sound recording unit 2. Silence is output for the frame time, and the recorded frame is reproduced as audio as it is.

FIG. 2 is a diagram showing transition of audio power in the embodiment. In FIG. 2, the portion where the power exceeds the voiced / silence determination value K is the voice data determined to be the voiced frame Fb. The portion where the power does not exceed the sound / silence determination value K is audio data determined to be a silent frame Fa.
A positive integer in the silence frame Fa before the speech frame Fb when changing from the silence frame Fa to the speech frame Fb
The monotonically increasing frame Fc in which the power continuously increases monotonically before the voiced frame Fb in the number of frames and the voiced frame after the voiced frame Fb are recorded in the voice recording unit 2 as voice data. is there.

[0015]

As is apparent from the above description, the voice storage device according to the first and second aspects temporarily stores the latest input voice data for a positive integer number of frames, and further stores the positive voice data. An increase or decrease in power adjacent to each of the integer number of frames is detected and stored. When the voice activity determination result is changed from a silence, based on the positive integer number of frames power fluctuation information stored, among the positive integer number of frames stored Since the voice data is recorded by adding only the frame whose power has been monotonically increased continuously before the voiced frame of to the beginning of a positive integer number of voiced frames, the beginning of the speech is like a Japanese syllabary. In the case where the voice power is monotonically increasing, the voice power is small, so that it is possible to prevent the speech head from being cut off by a frame determined to be silent.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of a voice storage device of the present invention.

FIG. 2 is an explanatory diagram showing transition of audio power in the embodiment.

FIG. 3 is a block diagram showing a conventional voice storage device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Silence detection part 2 Audio recording part 3 Audio reproduction part 4 Control part 5 Frame storage part 6 Power increase / decrease detection part Sa Input audio signal Sb, Sc Audio signal Sd Output audio signal Se Sound / silence determination information Ca Audio recording control signal Cb Audio Reproduction control signal Sf Frame signal Sg Power increase / decrease information

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-64-76098 (JP, A) JP-A-4-63047 (JP, A) JP-A-62-108645 (JP, A) 164899 (JP, U) (58) Fields investigated (Int. Cl. ⁷ , DB name) G10L 3/00 G10L 9/18 H04M 1/64-1/658

Claims (2)

(57) [Claims] 1. A sound / silence determining means for detecting the power of digitally encoded audio data to determine whether the audio data is audio or non-sound, and a latest positive integer number of frames of the digitally encoded audio data. And a positive integer stored in the frame storage means.
Power increase / decrease detecting means for detecting and storing power increase / decrease with respect to each adjacent frame for the number of frames, and a case where the frame changes from a silent frame to a sound frame based on the determination by the voice / silence determining means. , Based on the power increase / decrease information for the positive integer number of frames stored in the power increase / decrease detection means, and consecutively before the sound frame of the positive integer number of frames stored in the frame storage means. And a sound recording unit that adds only a frame whose power monotonically increases to the beginning of the sound frame and records the frame in the sound storage unit. 2. The sound / silence determining means determines the sound data when the power exceeds the sound / silence determination value as a sound frame, determines the sound data when the power does not exceed the sound / non-sound frame, and stores the frame as a soundless frame. A monotonically increasing frame in which the power is monotonically increasing continuously before the voiced frame in a positive integer number of frames in the voiceless frame before the voiced frame when the means changes from a voiceless frame to a voiced frame, 2. The voice storage device according to claim 1, wherein a voice frame after the voice frame is recorded as voice data.