JPH09113350A - Method and apparatus for predicting average level of background noise - Google Patents

Method and apparatus for predicting average level of background noise

Info

Publication number
JPH09113350A
JPH09113350A JP26573095A JP26573095A JPH09113350A JP H09113350 A JPH09113350 A JP H09113350A JP 26573095 A JP26573095 A JP 26573095A JP 26573095 A JP26573095 A JP 26573095A JP H09113350 A JPH09113350 A JP H09113350A
Authority
JP
Japan
Prior art keywords
section
level
background noise
time
average level
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
JP26573095A
Other languages
Japanese (ja)
Other versions
JP3244252B2 (en
Inventor
Yoichi Haneda
陽一 羽田
Shoji Makino
昭二 牧野
Masafumi Tanaka
雅史 田中
Jiyunko Shimizu
潤子 清水
Junji Kojima
順治 小島
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nippon Telegraph and Telephone Corp
Original Assignee
Nippon Telegraph and Telephone Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nippon Telegraph and Telephone Corp filed Critical Nippon Telegraph and Telephone Corp
Priority to JP26573095A priority Critical patent/JP3244252B2/en
Publication of JPH09113350A publication Critical patent/JPH09113350A/en
Application granted granted Critical
Publication of JP3244252B2 publication Critical patent/JP3244252B2/en
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Landscapes

  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
  • Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)

Abstract

PROBLEM TO BE SOLVED: To determine the average level of background noise by dividing the range of background noise level into a plurality of sections, determining the generation frequency of short time average level obtained from a sound receiving signal, and then judging a long time average level obtained independently based on a threshold value corresponding to a value in the minimum section among peak sections. SOLUTION: Output from a microphone 10 is passed through an A/D converter 11 and fed to power calculating sections 12, 13 where short time and long time root mean powers are determined. A histogram calculating section 14 determines the histogram of short time root mean power and an adaptive threshold determining section 15 determines a threshold value corresponding to the value in minimum section among the peak section of frequency of occurrence. The long time root mean power is compared with its threshold value and the number of times when the long time root mean power is consecutively lower than the threshold value is counted at a counting section 17. When a predetermined count is reached, a long time root mean power is outputted from a background noise mean amplitude determining section 18 thus determining an accurate mean amplitude of background noise.

Description

【発明の詳細な説明】Detailed Description of the Invention

【0001】[0001]

【発明の属する技術分野】この発明は、受音器で受音さ
れた信号の中から背景雑音の平均レベル(平均パワー又
は平均振幅値)を予測し、例えば背景雑音の混入した音
声信号からの音声信号検出や、2線4線変換系および拡
声通話系などで使用される反響消去装置での背景雑音平
均振幅値を予測するのに適用される背景雑音平均レベル
予測方法及び装置に関するものである。
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention predicts an average level (average power or average amplitude value) of background noise from a signal received by a sound receiver and, for example, from an audio signal mixed with background noise. BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a background noise average level prediction method and apparatus applied to predict a background noise average amplitude value in an echo canceller used in voice signal detection, a two-wire four-wire conversion system, a voice communication system, and the like. .

【0002】[0002]

【従来の技術】背景雑音平均振幅値の2乗である背景雑
音平均レベルを正確に予測することは、音声検出技術や
音声スイッチ、反響消去装置などの音情報装置において
大変重要な技術である。一般にマイクロホンを使用して
音声を受音した場合には、常に空調音などの背景雑音が
混入する。この時、従来においては音声の有音区間のみ
を検出しようとした場合に、あらかじめ予測される背景
雑音レベルをしきい値P t とし、マイクロホン受音信号
のレベルPx に対し、Px >Pt であれば有音区間であ
ると判断するような制御が行われる。しかし、背景雑音
があらかじめ決めたしきい値以上であれば、常に有音区
間であると判断されてしまうという問題があった。
Background noise Background noise which is the square of the average amplitude value of background noise.
Accurate prediction of the average sound level depends on voice detection technology and
In sound information devices such as voice switches and echo cancellers
This is a very important technology. Generally using a microphone
When receiving sound, background noise such as air conditioning noise is always
mixing. At this time, conventionally only the voiced section of the voice
Predictive background when trying to detect
Noise level is threshold P tAnd microphone received signal
Level PxOn the other hand, Px> PtIf it is a sound section
The control is performed to determine that But the background noise
Is equal to or greater than a predetermined threshold,
There was a problem that it was decided that it was in the middle.

【0003】そこで、背景雑音レベルを適応的に予測
し、この予測値に基づいてしきい値を決定する方法が提
案されている。この方法としては、マイクロホン受音信
号を一定時間区間毎に区切り、各時間区間の平均パワー
を求め、これをある時間分だけメモリに蓄積し、その平
均パワーの中から最小のものを背景雑音レベルとする方
法がある。しかしながら、この方法では、瞬間的に背景
雑音が小さくなった場合などで背景雑音レベルが実際よ
り小さく予測されるといった問題が生じる。
Therefore, a method has been proposed in which the background noise level is adaptively predicted and the threshold value is determined based on this predicted value. This method divides the microphone sound signal into fixed time intervals, calculates the average power of each time interval, stores this for a certain amount of time in memory, and selects the minimum of the average power as the background noise level. There is a way to. However, this method has a problem that the background noise level is predicted to be smaller than the actual value when the background noise is momentarily reduced.

【0004】また、各時間区間毎の平均パワーのレベル
差が、背景雑音では音声に比べ小さいことに着目し、時
間区間毎の平均パワーのレベル差が一定値以下の場合に
は、その平均パワーを背景雑音レベルとする方法もあ
る。この方法においても、時間区間の平均パワーのレベ
ル差がどの程度になれば良いかを試行錯誤的に決める必
要があり、汎用性に欠ける。
Also, paying attention to the fact that the level difference of the average power in each time section is smaller than that of the voice in the background noise, and when the level difference of the average power in each time section is less than a certain value, the average power is There is also a method of using as the background noise level. Also in this method, it is necessary to determine by trial and error how much the level difference of the average power in the time section should be, and this lacks versatility.

【0005】[0005]

【発明が解決しようとする課題】従来、背景雑音のレベ
ルを予測する方法として決められた時間区間毎の平均パ
ワーレベルを用いて行う方法では、時間区間毎の平均パ
ワーの最小値を使用すると、予測される背景雑音レベル
が小さくなり過ぎ、またあらかじめ決めたしきい値を用
いると、汎用性に欠けるといった問題があった。この発
明は、情報信号に混入している背景雑音のレベルを音声
などの情報信号と背景雑音が常に混在している場合にお
いても、汎用的に予測することができる方法及び装置を
提供することを目的としている。
In the conventional method of using the average power level for each time interval determined as a method for predicting the level of background noise, if the minimum value of the average power for each time interval is used, There is a problem that the predicted background noise level becomes too small, and if a predetermined threshold value is used, it lacks versatility. The present invention provides a method and apparatus that can predict the level of background noise mixed in an information signal in a general-purpose manner even when the information signal such as voice and the background noise are always mixed. Has an aim.

【0006】[0006]

【課題を解決するための手段】この発明では、マイクロ
ホンで受音された信号の短時間平均レベルをその短時間
ごとに計算し、背景雑音レベルの範囲を複数の区間毎に
分け、その各区間について計算した短時間平均レベルの
発生頻度、つまりヒストグラムを計算し、そのピークと
なる区間のうち最小の区間のレベルと対応する値を適応
的しきい値とし、前記短時間よりも長い時間における受
音信号の平均レベルを求め、この長時間平均レベルが前
記適応的しきい値以下に一定時間連続して存在する場合
には、その長時間平均レベル値を背景雑音平均レベルと
する。
According to the present invention, a short-time average level of a signal received by a microphone is calculated for each short time, the background noise level range is divided into a plurality of sections, and each of the sections is divided. The frequency of occurrence of the short-time average level calculated for, that is, a histogram is calculated, and the value corresponding to the level of the minimum section of the peak section is used as an adaptive threshold, and the threshold value for a time longer than the short section is received. The average level of the sound signal is obtained, and if the long-term average level is continuously lower than the adaptive threshold value for a certain period of time, the long-time average level value is set as the background noise average level.

【0007】さらに、背景雑音レベルの変化に追従する
ために、過去のヒストグラムデータに対し忘却係数を前
記短時間平均レベルの計算ごとに乗算する。この構成に
よれば、短時間平均レベルのヒストグラムを用いている
ために瞬間的に小さくなるような背景雑音に対してもそ
の瞬間低下に影響されることなく、平均値を確実に求め
ることが可能となる。
Further, in order to follow the change in the background noise level, the forgetting factor is multiplied to the past histogram data for each calculation of the short-time average level. According to this configuration, since the histogram of the short-time average level is used, it is possible to reliably obtain the average value without being affected by the instantaneous decrease even for the background noise that becomes momentarily small. Becomes

【0008】[0008]

【発明の実施の形態】図1にこの発明による装置の実施
例を示す。マイクロホン10の出力はAD変換器11で
例えば1/(8kHz)周期でデジタル信号とされて短
時間パワー計算部12及び長時間パワー計算部13でそ
れぞれ短時間平均パワーの平方根及び長時間平均パワー
の平方根が計算され、その短時間平均パワー平方根はそ
のヒストグラムがヒストグラム計算部14で計算され、
そのヒストグラムにもとづいてしきい値Pt が適応的し
きい値決定部15で決定され、そのしきい値Pt と長時
間平均パワー平方根値とが比較部16で比較され、その
比較結果の連続してしきい値以下の回数が計数部17で
計数され、その計数値が所定値になるとその時の長時間
平均パワー平方根値が背景雑音平均振幅値決定部18か
ら出力される。
1 shows an embodiment of the device according to the present invention. The output of the microphone 10 is converted into a digital signal with a period of 1 / (8 kHz) by the AD converter 11, and the short-time power calculation unit 12 and the long-time power calculation unit 13 respectively calculate the square root of the short-time average power and the long-time average power. The square root is calculated, and the histogram of the short-time average power square root is calculated by the histogram calculation unit 14,
The threshold value P t is determined by the adaptive threshold value determination unit 15 based on the histogram, the threshold value P t and the long-time average power square root value are compared by the comparison unit 16, and the comparison results are continuously calculated. Then, the number of times equal to or less than the threshold value is counted by the counting unit 17, and when the count value reaches a predetermined value, the long-time average power square root value at that time is output from the background noise average amplitude value determination unit 18.

【0009】図1に示した装置を用いるこの発明による
背景雑音平均レベル推定方法の処理手順を示す。図1及
び図2を参照してこの発明による装置の動作と、この発
明の方法とを説明する。まず計数部17の計数値Kを0
に初期化し(S1 )、マイクロホン10により受音さ
れ、AD変換器11でデジタル信号とされた受音信号x
(n)(nはサンプリング時刻を示す)は短時間パワー
計算部12及び長時間パワー計算部13へ供給され予め
決められた短時間、例えば5〜10ミリ秒間(これを長
くすると背景雑音と音声との区別ができなくなる)にお
ける平均パワーP s(n)及び予め決められた長時間、
例えば100〜300ミリ秒間(長いと平均精度はよく
なるが音声信号も混入してしまう)における平均パワー
L(n)がそれぞれ次式で計算される。
According to the invention using the device shown in FIG.
The processing procedure of the background noise average level estimation method is shown. Figure 1 and
And the operation of the device according to the invention and
Ming's method is explained. First, the count value K of the counter 17 is set to 0.
Initialized to (S1), Received by the microphone 10
And the sound reception signal x converted into a digital signal by the AD converter 11
(N) (n indicates sampling time) is short-time power
It is supplied to the calculation unit 12 and the long-term power calculation unit 13 in advance.
A fixed short time, for example 5-10 milliseconds
If you turn it off, you will not be able to distinguish the background noise from the voice).
Average power P s(N) and a predetermined long time,
For example, 100 to 300 milliseconds (the longer the average accuracy, the better
However, the audio signal is also mixed in)
PLEach (n) is calculated by the following equation.

【0010】 PS(n)=α1 ×PS(n−1)+(1−α1 )×x2(n) (1) PL(n)=α2 ×PL(n−1)+(1−α2 )×x2(n) (2) α1 ,α2 はそれぞれパワー平均の時間に関係する1以
下の量であり、積分時間(平均時間)をTとするとT=
1/(1−α)の関係が成り立つ。従ってα1<α2
1となる。更にこれら平均パワーPS(n),PL(n)
の平方根√PS(n)、つまり振幅、√PL(n)、つま
り平均振幅を求める(S2 ,S3)。これらの演算はn
ごと、つまり受音信号のサンプリング周期ごとに行われ
る。
P S (n) = α 1 × P S (n-1) + (1-α 1 ) × x 2 (n) (1) P L (n) = α 2 × P L (n-1) ) + (1−α 2 ) × x 2 (n) (2) α 1 and α 2 are quantities less than or equal to 1 related to the power averaging time, and T = T when the integration time (averaging time) is T =
The relationship of 1 / (1-α) is established. Therefore α 12 <
It becomes 1. Further, these average powers P S (n) and P L (n)
The square root √P S (n), that is, the amplitude, √P L (n), that is, the average amplitude is obtained (S 2 , S 3 ). These operations are n
Every time, that is, every sampling period of the sound reception signal.

【0011】次にヒストグラム計算部14において短時
間パワー平方根√PS(n)のヒストグラムを計算する
(S4)。つまり予測される√PS(n)の範囲、例えば
最大音声振幅の半分(背景雑音は通常、音声より可成小
さい)を等間隔δで複数の振幅区間(0,1,…,N)
に分割し、短時間パワー平方根√PS(n)を計算する
ごとに、その値が属する振幅区間の数を1加算する。つ
まり下記演算を行う。
Next, the histogram calculation unit 14 calculates a histogram of the short-time power square root √P S (n) (S 4 ). That is, a predicted range of √P S (n), for example, half of the maximum voice amplitude (background noise is usually smaller than voice) is divided into a plurality of amplitude sections (0, 1, ..., N) at equal intervals δ.
Every time the short-time power square root √P S (n) is calculated, the number of amplitude sections to which the value belongs is incremented by one. That is, the following calculation is performed.

【0012】 h(int(√PS(n)/δ))=h(int(√PS(n)/δ))+1 …(3) int(A)はAの値の小数点以下を切捨て整数値化す
ることを示す。従って例えばδ=40で、√PS(n)
が50であればh( int(50/40))=h(1)とな
り、1番目の振幅区間に1を加算し、また√PS(n)
が90であれば、h( int(90/40))=h(2)と
なり、2番目の振幅区間に1を加算する。
H (int (√P S (n) / δ)) = h (int (√P S (n) / δ)) + 1 (3) int (A) rounds down the fractional part of the value of A Indicates that the value is an integer. Therefore, for example, when δ = 40, √P S (n)
Is 50, h (int (50/40)) = h (1), 1 is added to the first amplitude section, and √P S (n)
Is 90, h (int (90/40)) = h (2), and 1 is added to the second amplitude section.

【0013】このヒストグラムを背景雑音レベルの変化
に追従しやすくするために、1以下の忘却係数λ(例え
ば500〜1000ミリ秒程度)を、次式に示すよう
に、各振幅区間0〜Nの個数にそれぞれ乗算する
(S5 )。 h(i)=λ×h(i) …(4) i=0,1,2,…,N 次に、計算されたヒストグラムh(i)(i=0,…,
N)は適応的しきい値決定部15に転送される。適応的
しきい値決定部15では、h(i)(i=0,…,N)
に対し、ピークとなっているh(i)を検索し、つまり
各区間の出現頻度h(i)についてその前後の区間の出
現頻度h(i−1)、h(i+1)より大きい振幅区間
を探し、その各区間番号を記憶する(S6 )。さらにそ
の検索したピークの振幅区間中で最小である区間番号i
を求め、これをwとする(S7 )。音声と背景雑音が混
在した信号が図3Aに示す場合に、その短時間パワー平
方根のヒストグラムh(i)が図3Bに示すようになっ
た場合、第1番目の振幅区間の出現率h(1)がピーク
となっていて、かつその区間番号iが最小であって、w
=1となる。その結果、適応的しきい値Pt (n)を、
int( )関数による切り捨ての効果を考慮して、下
記式により求める(S8 )。
In order to make this histogram easy to follow changes in the background noise level, a forgetting factor λ of 1 or less (for example, about 500 to 1000 milliseconds) is set for each amplitude section 0 to N as shown in the following equation. Each number is multiplied (S 5 ). h (i) = λ × h (i) (4) i = 0, 1, 2, ..., N Next, the calculated histogram h (i) (i = 0, ..., N).
N) is transferred to the adaptive threshold value determining unit 15. In the adaptive threshold value determining unit 15, h (i) (i = 0, ..., N)
On the other hand, the peak h (i) is searched for, that is, with respect to the appearance frequency h (i) of each section, an amplitude section larger than the appearance frequencies h (i-1) and h (i + 1) of the sections before and after that is found. looking, and stores the each section number (S 6). Further, the section number i that is the smallest in the amplitude section of the retrieved peak
Is obtained and is set as w (S 7 ). When the signal in which voice and background noise are mixed is shown in FIG. 3A and the histogram h (i) of the short-time power square root thereof is as shown in FIG. 3B, the appearance rate h (1 ) Is a peak, and the section number i is the smallest, and w
= 1. As a result, the adaptive threshold P t (n) is
In consideration of the effect of truncation by the int () function, it is determined by the following formula (S 8 ).

【0014】 Pt(n)=δ×(w+1) …(5) ここで、実際の信号を例に取り、適応的しきい値の決定
についてさらに説明を加える。図3Aに、マイクロホン
10で観測された信号x(n)を示す。x(n)は、背
景雑音に音声信号が加算された信号である。図3Bに、
振幅区間δを40としてヒストグラム計算部14で計数
したヒストグラムを示す。適応的しきい値決定部15で
は、ピークを示す出現頻度の内で、最小の振幅区間番号
を持つものを検索するが、これは図3Bから区間1であ
ることが分かる。その結果、適応的しきい値は、(5)
式に従い、 Pt(n)=40×(1+1)=80 …(6) と計算される。これは振幅絶対値40〜80の出現頻度
の振幅区間番号1がピーク区間の最小区間番号であるこ
とに対応している。
P t (n) = δ × (w + 1) (5) Here, taking an actual signal as an example, the determination of the adaptive threshold value will be further described. FIG. 3A shows a signal x (n) observed by the microphone 10. x (n) is a signal in which the voice signal is added to the background noise. In FIG. 3B,
The histogram calculated by the histogram calculation unit 14 when the amplitude section δ is 40 is shown. The adaptive threshold value determining unit 15 searches for the one having the smallest amplitude section number among the appearance frequencies showing the peaks, and it can be seen from FIG. 3B that it is the section 1. As a result, the adaptive threshold is (5)
According to the formula, P t (n) = 40 × (1 + 1) = 80 (6) is calculated. This corresponds to that the amplitude section number 1 of the appearance frequency of the absolute amplitude values 40 to 80 is the minimum section number of the peak section.

【0015】比較部16では、適応的しきい値決定部1
6で決定された適応的しきい値Pt(n)と、長時間パワ
ー計算部13により計算された長時間平均パワー平方根
L (n)とを比較し(S9)、PL(n)<Pt(n)で
あれば、計数部17の計数値kを1加算し(S10)、こ
の計数値が所定値K以上になったかを判定する(S1 1)
。比較部16でPL(n)がPt(n)以下であれば計
数部17の計数値kを0にリセットしてステップS2
戻り(S12)、ステップS11でk<Kであればステップ
2 に戻る。ステップS11でk>Kならばつまり所定時間
連続してPL(n)<Pt(n)であれば、PL(n)を背
景雑音平均振幅値決定部18は背景雑音の平均振幅値と
決定する( S13)。Kの値は時間にして例えば1秒程
度、つまり相互の会話が明らかに途切れている時間とさ
れる。
In the comparison unit 16, the adaptive threshold value determination unit 1
Adaptive threshold P determined in 6t(n) and long time power
-Long time average power square root calculated by the calculator 13.
PL Compare with (n) (S9), PL(N) <PtIn (n)
If there is, the count value k of the counting unit 17 is incremented by 1 (STen), This
It is judged whether the count value of is equal to or more than the predetermined value K (S1 1)
 . P in the comparison unit 16L(N) is PtIf (n) or less, total
The count value k of the number part 17 is reset to 0 and step STwoTo
Return (S12), Step S11If k <K, step
TwoReturn to Step S11If k> K, that is, predetermined time
P in successionL(n) <PtIf (n), then PL(N) back
The background noise average amplitude value determination unit 18 calculates the average amplitude value of background noise and
Determine (S13). The value of K is about 1 second in time
Degree, that is, the time and
It is.

【0016】図2における各処理は前述したように受音
信号のサンプリング周期で行われる。長時間パワー計算
部13では、初期状態では、それまでのPL(n)がゼロ
とされ、平均のための長時間分の時間が経過した後、こ
の発明による処理が正しく行われる。上述の実施例では
平均パワーの平方根をとり平均振幅を予測したが、上述
の処理で平方根をとることなく、平均パワーで処理し、
背景雑音の平均パワーを予測してもよい。つまりこの発
明でレベルとは振幅又はパワーを意味するものである。
Each processing in FIG. 2 is performed at the sampling cycle of the sound reception signal as described above. In the long-time power calculation unit 13, in the initial state, P L (n) up to that point is set to zero, and after the time corresponding to the long time for averaging has elapsed, the process according to the present invention is correctly performed. In the above-mentioned embodiment, the square root of the average power was taken to predict the average amplitude, but the square root was not taken in the above processing, but the average power was processed,
The average power of background noise may be predicted. That is, in the present invention, the level means amplitude or power.

【0017】[0017]

【発明の効果】以上説明したように、この発明によれば
事前に背景雑音レベルに仮定を置くことなく、音声と背
景雑音が混在している場合においても、適応的に背景雑
音平均レベル値を適応的しきい値として予測することが
可能である。さらにこの適応的しきい値以下である正確
な背景雑音平均レベル値を測定することができる。この
発明は正確な背景雑音平均レベル値が測定可能であり、
音声検出やノイズリダクション等に利用することによ
り、音声検出の高精度化やノイズ消去量の改善が図れ
る。
As described above, according to the present invention, the background noise average level value is adaptively set even when the voice and the background noise are mixed without making an assumption on the background noise level in advance. It can be predicted as an adaptive threshold. Furthermore, it is possible to measure an accurate background noise average level value that is below this adaptive threshold. This invention can measure accurate background noise average level value,
By using it for voice detection, noise reduction, etc., it is possible to improve the accuracy of voice detection and improve the amount of noise elimination.

【図面の簡単な説明】[Brief description of the drawings]

【図1】この発明装置の実施例を示すブロック図。FIG. 1 is a block diagram showing an embodiment of the device of the present invention.

【図2】この発明方法の実施例を示す流れ図。FIG. 2 is a flow chart showing an embodiment of the method of the present invention.

【図3】Aはこの発明の効果を示すために使用した音声
と背景雑音が混在した信号を示す図、Bはこの発明の中
のヒストグラム計算部14の計算結果を示す図である。
3A is a diagram showing a signal mixed with voice and background noise used to show the effect of the present invention, and FIG. 3B is a diagram showing a calculation result of a histogram calculating section 14 in the present invention.

───────────────────────────────────────────────────── フロントページの続き (72)発明者 清水 潤子 東京都千代田区内幸町1丁目1番6号 日 本電信電話株式会社内 (72)発明者 小島 順治 東京都千代田区内幸町1丁目1番6号 日 本電信電話株式会社内 ─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Junko Shimizu 1-1-6 Uchisaiwaicho, Chiyoda-ku, Tokyo Nihon Telegraph and Telephone Corp. Nippon Telegraph and Telephone Corporation

Claims (3)

【特許請求の範囲】[Claims] 【請求項1】 所定の短時間の平均入力信号のレベルを
求める段階と、 背景雑音レベルの所定の範囲を複数に分割した各区間に
ついて上記求めた短時間平均レベルが出現する頻度を求
める段階と、 上記各区間について、その前後の区間における上記出現
頻度よりも出現頻度が大となっているピーク区間中の最
小区間の区間値を求める段階と、 上記短時間よりも長い時間の上記入力信号の平均レベル
を求める段階と、 上記最小区間の区間値と対応したレベルと上記長時間の
平均レベルとを比較する段階と、 その比較で長時間平均レベルの方が小さいと判定される
ことが連続して所定時間得られると上記長時間平均レベ
ルを背景雑音の平均レベルとする段階と、 を有する背景雑音平均レベル予測方法。
1. A step of obtaining a level of an average input signal for a predetermined short time, and a step of obtaining a frequency at which the above-obtained short-time average level appears in each section obtained by dividing a predetermined range of the background noise level into a plurality of sections. For each section, the step of obtaining the section value of the minimum section in the peak section whose appearance frequency is higher than the appearance frequency in the sections before and after the section, and the input signal of the time longer than the short time The step of obtaining the average level, the step of comparing the level corresponding to the section value of the minimum section with the average level of the long time, and the comparison that it is judged that the long-term average level is smaller is continuous. For a predetermined time, the average level of the long-term average is used as the average level of background noise, and the average level of background noise is predicted.
【請求項2】 上記ピーク区間の最小区間を求めること
を、各区間の出現頻度に忘却係数を乗算した後行うこと
を特徴とする請求項1記載の背景雑音平均レベル予測方
法。
2. The background noise average level prediction method according to claim 1, wherein the minimum section of the peak section is obtained after multiplying the appearance frequency of each section by a forgetting factor.
【請求項3】 到来音を受音するマイクロホンと、 上記マイクロホンの出力の所定の短時間の平均レベルを
求める短時間レベル計算手段と、 上記マイクロホンの出力の、上記短時間よりも長い時間
における平均レベルを求める長時間レベル計算手段と、 背景雑音レベルの所定の範囲を複数に分割した各区間に
ついて、上記求めた短時間平均レベルが出現する頻度を
求めるヒストグラム計算手段と、 上記各区間についてその前後の区間における出現頻度よ
りも出現頻度が大となっているピーク区間中の最小区間
の区間値と対応したレベルをしきい値として求める適応
的しきい値決定手段と、 上記しきい値と上記長時間平均レベルとを比較する比較
手段と、 上記比較においてしきい値以下が所定回数連続して得ら
れると、その時の長時間平均レベルを背景雑音レベルと
して出力する背景雑音平均レベル決定手段と、 を具備する背景雑音平均レベル予測装置。
3. A microphone for receiving an incoming sound, a short time level calculating means for obtaining an average level of the output of the microphone for a predetermined short time, and an average of the output of the microphone for a time longer than the short time. Long-time level calculating means for obtaining the level, histogram calculating means for obtaining the frequency at which the obtained short-time average level appears for each section obtained by dividing the predetermined range of the background noise level into a plurality of sections, and before and after each section. An adaptive threshold value determining means for obtaining as a threshold value a level corresponding to the interval value of the minimum interval in the peak interval in which the appearance frequency is higher than the appearance frequency in the interval, the threshold value and the length When the comparison means for comparing the time average level with the time average level is continuously obtained a predetermined number of times or less in the above comparison, the long time average level at that time is obtained. A background noise average level predicting apparatus comprising: a background noise average level determining means for outputting a bell as a background noise level.
JP26573095A 1995-10-13 1995-10-13 Background noise average level prediction method and apparatus Expired - Lifetime JP3244252B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP26573095A JP3244252B2 (en) 1995-10-13 1995-10-13 Background noise average level prediction method and apparatus

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP26573095A JP3244252B2 (en) 1995-10-13 1995-10-13 Background noise average level prediction method and apparatus

Publications (2)

Publication Number Publication Date
JPH09113350A true JPH09113350A (en) 1997-05-02
JP3244252B2 JP3244252B2 (en) 2002-01-07

Family

ID=17421207

Family Applications (1)

Application Number Title Priority Date Filing Date
JP26573095A Expired - Lifetime JP3244252B2 (en) 1995-10-13 1995-10-13 Background noise average level prediction method and apparatus

Country Status (1)

Country Link
JP (1) JP3244252B2 (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO1999041848A1 (en) * 1998-02-13 1999-08-19 Telefonaktiebolaget Lm Ericsson (Publ) Methods and apparatus for controlling filter adaption in noise
JP2007065832A (en) * 2005-08-30 2007-03-15 Aisin Seiki Co Ltd Exterior noise processor
WO2009096551A1 (en) * 2008-01-30 2009-08-06 Jfe Advantech Co., Ltd. Diagnostic system for bearing
JP2017530357A (en) * 2014-09-24 2017-10-12 ローズマウント インコーポレイテッド Acoustic detection in process environments

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO1999041848A1 (en) * 1998-02-13 1999-08-19 Telefonaktiebolaget Lm Ericsson (Publ) Methods and apparatus for controlling filter adaption in noise
JP2007065832A (en) * 2005-08-30 2007-03-15 Aisin Seiki Co Ltd Exterior noise processor
WO2009096551A1 (en) * 2008-01-30 2009-08-06 Jfe Advantech Co., Ltd. Diagnostic system for bearing
JPWO2009096551A1 (en) * 2008-01-30 2011-05-26 Jfeアドバンテック株式会社 Bearing diagnosis system
JP4874406B2 (en) * 2008-01-30 2012-02-15 Jfeアドバンテック株式会社 Bearing diagnosis system
KR101429952B1 (en) * 2008-01-30 2014-08-13 제이에프이 어드밴테크 가부시키가이샤 Diagnostic system for bearing
JP2017530357A (en) * 2014-09-24 2017-10-12 ローズマウント インコーポレイテッド Acoustic detection in process environments
US10228351B2 (en) 2014-09-24 2019-03-12 Rosemount Inc. Acoustic detection in process environments

Also Published As

Publication number Publication date
JP3244252B2 (en) 2002-01-07

Similar Documents

Publication Publication Date Title
CN101826892B (en) Echo canceller
US5570423A (en) Method of providing adaptive echo cancellation
EP0790599B1 (en) A noise suppressor and method for suppressing background noise in noisy speech, and a mobile station
CN101321201B (en) Echo elimination device, communication terminal and method for confirming echo delay time
US6990194B2 (en) Dynamic balance control for telephone
US4571461A (en) Conference telephone apparatus
JP2573352B2 (en) Voice detection device
CN1331883A (en) Methods and appts. for adaptive signal gain control in communications systems
US8565414B2 (en) Distributed VAD control system for telephone
JPH07193520A (en) Equipment and method to discriminate noise in input signal
CA2309525C (en) Method of detecting silence in a packetized voice stream
JPH09113350A (en) Method and apparatus for predicting average level of background noise
JP2919685B2 (en) Signal identification circuit
US6842526B2 (en) Adaptive noise level estimator
EP1229517B1 (en) Method for recognizing speech with noise-dependent variance normalization
KR100386485B1 (en) Transmission system with improved sound
US6754337B2 (en) Telephone having four VAD circuits
CN114071220A (en) Sound effect adjusting method and device, storage medium and electronic equipment
JP3380048B2 (en) Fading pitch estimation device
US6847930B2 (en) Analog voice activity detector for telephone
CN101727913A (en) Noise estimation apparatus, calling apparatus, and noise estimation method
JPH05183997A (en) Automatic discriminating device with effective sound
JP3678611B2 (en) Audio conference system
JPH08223108A (en) Fading pitch estimating device
JP2949925B2 (en) Sound effect addition discriminator

Legal Events

Date Code Title Description
FPAY Renewal fee payment (prs date is renewal date of database)

Free format text: PAYMENT UNTIL: 20071026

Year of fee payment: 6

FPAY Renewal fee payment (prs date is renewal date of database)

Free format text: PAYMENT UNTIL: 20081026

Year of fee payment: 7

FPAY Renewal fee payment (prs date is renewal date of database)

Free format text: PAYMENT UNTIL: 20091026

Year of fee payment: 8

FPAY Renewal fee payment (prs date is renewal date of database)

Free format text: PAYMENT UNTIL: 20101026

Year of fee payment: 9

FPAY Renewal fee payment (prs date is renewal date of database)

Free format text: PAYMENT UNTIL: 20101026

Year of fee payment: 9

FPAY Renewal fee payment (prs date is renewal date of database)

Free format text: PAYMENT UNTIL: 20111026

Year of fee payment: 10

FPAY Renewal fee payment (prs date is renewal date of database)

Free format text: PAYMENT UNTIL: 20111026

Year of fee payment: 10

FPAY Renewal fee payment (prs date is renewal date of database)

Free format text: PAYMENT UNTIL: 20121026

Year of fee payment: 11

FPAY Renewal fee payment (prs date is renewal date of database)

Free format text: PAYMENT UNTIL: 20121026

Year of fee payment: 11

FPAY Renewal fee payment (prs date is renewal date of database)

Free format text: PAYMENT UNTIL: 20131026

Year of fee payment: 12

S531 Written request for registration of change of domicile

Free format text: JAPANESE INTERMEDIATE CODE: R313531

R350 Written notification of registration of transfer

Free format text: JAPANESE INTERMEDIATE CODE: R350

EXPY Cancellation because of completion of term