JPS5950499A - Sound-free compression control system - 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 The present invention (rl) reproduces digitized audio as @sound?'7 In a voice processing device, Freno of a sound part,
and the silent part of the phrase 1 that is written after the child's i'Jit.
Save only Ft! (L, Concerning a silence compression control method that makes it possible to compress the back part of a voice without making the beginning and end of a word unclear.

Since a normal conversation contains many silent parts, the memory capacity is reduced by recording only the vocal parts without recording the silent parts. However, when detecting the presence or absence of voice based on the power level of the input data, the silent part is mistakenly judged as the active part due to noise and recorded, resulting in scratches. For this reason,
(lY-East is the sound part and I, ・1(, f'J', gl
-Minute Threshold Sen☆)・,7,Prevents 4′I1 sound and sound. -:
The low beginning of the word and the ending of h are 4-・'l'%” F: Judging from the 1 minute and 1.
i J Koga 1.11 Ha “r3” with lost naturalness m
If I'm too lazy, I'll have two flaws.

The present invention was developed in view of the above-mentioned shortcomings.The audio to be digitized is not predetermined, but is divided into frames of length, and whether or not there is a sound is determined in units of frames. , Is the frame at the end of the frame?'71!
In addition to the blind 111 storage device that generates a flag indicating whether it is a u sound and stores only sound information, during recording, it is connected only to the frame of the sound part and the total frame of the 44+rγf part before and immediately after IC1. By storing the Freno number as ``1'' and ``F〆I'', the 874 beginning and ending of the sound are changed to ``kiruno 1''. To provide a sound compression control system capable of storing speech with compressed silent parts without losing natural +/1 ft.

Below JJ, the present invention will be explained based on the W, ffti example shown in the drawings.

FIG. 1:1: This is an embodiment of a configuration block diagram of the present invention. 1 is the input ψ of the digital audio data to be recorded
! i1: child, 2 is a voice detection circuit, which predetermines the digital voice data inputted through the input cuff 11 child 1; The length is divided into minutes (units) of 1jl.
Then, for each frame, it is determined whether there is a sound or a sound, and then the frame number is added to each frame and sent to the buffer 4, and then the frame of Guo Gong is determined. , the silence flag is sent to the silence flag detection circuit 3. The silence flag detection circuit 3 sends its output to a buffer 4 (c), and the audio detection circuit 2 sends its output to a buffer (c).

control the voice data that comes, and
(including the number) is stored in the memory 5 in a compressed form ρl.

Said voice detection circuit 2. Is the operation by the silence flag Hiiragi output circuit 3゜buffer 4 turned on during recording? Ear voice waveform, γr,
1. +“During, nothing?” Check each state of the 1 flag with IR error 2nd
(Figure and shows the state of entry into the buffer - J Go 1) 7
This will be explained and explained based on Figure 3. ? :X' T” f・
1z output circuit 2 outputs audio data to each of 71 nodes 3 to J.
Minutes:! +'J shi, l for each frame a-j
lki Next frame number 0) to O◇woi [Add. Then, in order to complete buffer 4, the first 71 norms C of the sound parts a, b and the silent part of the old p1 data are l
Write 1 at 1it'i next time. However, following θζ, <the silent part is the silence flag detection time t? By the signal from 'i 3, Pi 31'&I IC (d, c - h f,
g −+ 1+, i is arrogant. As soon as 1 is 91 recorded, the frame becomes the 1] sound part, so J is next σ) At 1/S! It's full size.

In addition, in the example, the 1115th minute of the A1 sound part is considered to be the sound part, and the first and last 1
17 is not necessarily limited to one frame.
The mark of the p sound flag is ilH[1'L, the previous frame with the flag (I1.11eC',] the $ sound flag located at L-m d, frame l/-m C) is a silent part.

Next, 7 is a blank number detection circuit, which takes out the contents of the memory 5 through a buffer 6 and reads 1 "to output Vr, ゛,
M students are played in the order of 7-ken numbers.

8 i-, i d+++ sound pattern sending circuit outputs a silent signal corresponding to a silent part of the audio data to the gate 9.

The gate 9 is used to change and output the 15-■ bow from the buffer 6 and the silent Bakun sending times 1';' and 8 in accordance with the Jit command from the Noken number detection circuit 7. figure/
In other words, the no-ken number detection circuit 7 detects the data internal unit from the memory 5, and the no-ken number detecting circuit 7 detects that the no-ken number is ils F2.
%lJ, if so, control the gate 9 to send out the signal from the buffer 6 as it is, and if there is a missing number in the sequence number, judge this missing number frame to be the part of the year note. The gate 9 is controlled to send out the signal from the silent pattern sending circuit 8. Shig, - Gatsui, Gate 0 to 111 Cuff A: 1 (1) J <
, τth (・1th 41st shi 1 wo l, zuyou l 13);-
It becomes i data, and it is one person with 7j I-Arc.

In addition,・,su,:・1. ? 4 pattern sending 11J1 route 8
As for the output, '/, 1 whole furnace, 1゛τ(, r 脣, white bean' ((1) can be m-).

#U of 1 (to the present invention), 1 is de fzikulized and rc i<voice'5 (・)l/-ノ, is cut into 1, phrase 1, nosiwa 1/pel detection at the enthronement. Is there a sound in the room? ′
At 16 minutes and the silent part, 〒11 cut off, Yagura 7:? By treating both or one of the frames of the free part immediately after the front part as the i/+ part, FBI (i/+) can be used for power level detection etc. Even if the audio is FFn beans 1 [, 53 tails will not be cut off,
I'll give you 111 voices that are as clear as possible and have no loss of naturalness.
f! f:

The sequence number that increments sequentially for each frame is added by τ1, and the frame number and audio data are stored only for the 8th part of the frame. When playing, the frame number is i'J.
<#TT I, it is determined that it is a sound frame and the 4-year-old audio data is played back, and when the frame number is non-freezing, it is determined that it is a silent frame and a silent pattern is sent U)J
It is always effective to easily compress the silent part during earing by sending out nail sound data from the road.

[Brief explanation of drawings]

2J'r 1 rgl (・One embodiment of the configuration block diagram of the present invention, 2nd 1 is the audio waveform and audio data at the time of recording. - (Input state diagram, Figure 4 shows the reproduced voice data. 1. Input 'Fj: '+5 child 2... Voice detection circuit 3 Enemy, i, T flag detection circuit 4... Buffer - r5...Memory 6...Buffer 7...
・Sequence number detection circuit 8...Silent pattern sending circuit 9...Game) 10...Output terminal Applicant: Japan i Figure 4 111 Sown 1 abCnnnnhj 'n: i sound ha 0 turn Continuation of page 1 ■Applicant: Nippon Telegraph and Telephone Public Corporation ■Applicant: Fujitsu Limited 648-1015 Kamiodanaka, Nakahara-ku, Kawasaki-shi

Claims (1)

[Claims] (11 digitized flp is determined in advance)
Divide into frames of high quality, detect whether each freno is a sound part) or a non-sound part, and issue a flag at the end of the frame to indicate whether the rq frame is a sound part or a silent part.
E t, te What B sound that accumulates only d) Garbage accumulation f1'
1) In II, 1) the silent part immediately before the sound part or immediately after the sound part. ), the i number of frames are regarded as parts with sound, and L −U V′; fL′I−J・1
It is characterized by
1 (. (2) Digitalized Her again, the voice was predetermined and divided into 1 parts of I-His and each Freno, each with 7 parts or 7 parts. It detects whether the Rk frame is the police section, and adds a flag to the frame location section to indicate whether the Rk frame is the 41 section or the no I'1 section, and then outputs the sound section t1φ.
1 (in a voice storage device, a predetermined number of frames of the recorded portion, both immediately before and after the sound part, are regarded as sound parts and stored!) Silence compression control method with % characteristics.