JP2004126855A - Accident sound detection circuit - Google Patents

Accident sound detection circuit Download PDF

Info

Publication number
JP2004126855A
JP2004126855A JP2002288517A JP2002288517A JP2004126855A JP 2004126855 A JP2004126855 A JP 2004126855A JP 2002288517 A JP2002288517 A JP 2002288517A JP 2002288517 A JP2002288517 A JP 2002288517A JP 2004126855 A JP2004126855 A JP 2004126855A
Authority
JP
Japan
Prior art keywords
sound
pitch
fluctuation
predetermined
accident
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
JP2002288517A
Other languages
Japanese (ja)
Other versions
JP3895657B2 (en
Inventor
Toshimitsu Seri
瀬利 俊光
Noboru Okino
興野 登
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.)
Mitsubishi Electric Engineering Co Ltd
Original Assignee
Mitsubishi Electric Engineering Co Ltd
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 Mitsubishi Electric Engineering Co Ltd filed Critical Mitsubishi Electric Engineering Co Ltd
Priority to JP2002288517A priority Critical patent/JP3895657B2/en
Publication of JP2004126855A publication Critical patent/JP2004126855A/en
Application granted granted Critical
Publication of JP3895657B2 publication Critical patent/JP3895657B2/en
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Images

Landscapes

  • Traffic Control Systems (AREA)

Abstract

<P>PROBLEM TO BE SOLVED: To provide an accident sound detection circuit improved in the detecting ability of accident sound by enabling the discrimination of a sudden brake sound from a siren. <P>SOLUTION: This detection circuit comprises an accident sound reading means 10 for extracting a prescribed signal component having a prescribed level or more of fluctuation of signal level from a voice signal of traffic sound; a pitch fluctuation determination arithmetic means 30 for determining the magnitude of fluctuation ratio of a basic frequency based on the prescribed signal component; and a pitch fluctuation sound removing means 20 for removing the prescribed signal component when the fluctuation ratio is larger than a prescribed fluctuation ratio, and outputting the prescribed signal component as the accident sound when the fluctuation ratio is the prescribed fluctuation ratio or less. The pitch fluctuation determination arithmetic means 30 includes a wavelet conversion arithmetic means 31 for wavelet-transforming the prescribed signal component at prescribed intervals; a pitch arithmetic means 32 for calculating the basic frequency between peaks of a prescribed value or more; and a pitch fluctuation determination means 33 for comparing the prescribed fluctuation ratio with the fluctuation ratio of the basic frequency to determine the magnitude of the fluctuation ratio. <P>COPYRIGHT: (C)2004,JPO

Description

【0001】
【発明の属する技術分野】
この発明は、交差点などにおける交通音から、事故音(急ブレーキ音など)のみを効率良く検出するための事故音検出回路に関するものである。
【0002】
【従来の技術】
一般に、交差点では交通事故が多く発生しており、従来から事故発生の状況を記録するための交通事故状況記録装置が存在している。交通事故状況記録装置には、周囲の音から事故音のみを検出するための事故音検出回路が備えられており、様々な方法が提案されている。
図9は、従来の交通事故状況記録装置を示すブロック図である。
図9において1、4は交通信号機、1B、1Y、1R、4B、4Y、4Rは信号灯、2B、2Y、2Rは信号線、3は信号灯点滅装置、5は信号灯記号発生装置、6は時刻文字発生部、7は加算部、8は撮像装置、9はエンドレス記録装置、10は周囲音検出装置、11は衝突音検出回路、12は記録制御装置、13は磁気記録再生装置である。
【0003】
周囲音検出装置10は、交差点周囲の音(交通音)を検出し、交通音に対応した音声信号を衝突音検出回路11に送信する。衝突音検出回路11は、交通音の音声信号の信号レベルと閾値とを比較し、信号レベルが閾値以上である場合に、車両の衝突音(事故音)であると判定する。
記録制御装置12は、交通音から事故音が検出された場合に、エンドレス記録装置に記録された交差点の映像などを再生させる。
このように、従来の事故音検出回路は、交通音の音声信号の信号レベルと閾値とを比較することによって事故音を検出している(例えば、特許文献1参照)。
【0004】
【特許文献1】
特開平4−338900号公報(第4頁、第2図)
【0005】
【発明が解決しようとする課題】
従来の事故音検出回路は以上のように、検出された音声信号の信号レベルと、閾値との比較により事故音を判定しているが、一般の交通音では例えばサイレン音など、事故音以外でもレベルの大きな音が多く存在している。そのため、このような検出された音声信号と閾値とのレベル比較を行う方法では、急ブレーキ音などの実際の事故音に加えて、サイレン音も事故音と判定してしまうという問題点があった。
【0006】
この発明は上記のような問題点を解決するためになされたもので、急ブレーキ音とサイレン音との識別を可能とし、事故音の検出能力を向上させることのできる事故音検出回路を得ることを目的とする。
【0007】
【課題を解決するための手段】
この発明に係る事故音検出回路は、交通音の音声信号から、信号レベルの変動量が所定レベル以上の所定信号成分を抽出する事故音取込手段と、所定信号成分により基本周波数の変動率の大小を判定するピッチ変動判定演算手段と、変動率が所定変動率よりも大きい場合、所定信号成分を除去し、変動率が所定変動率以下の場合、所定信号成分を事故音として出力するピッチ変動音除去手段とを備え、ピッチ変動判定演算手段は、所定信号成分を所定間隔でウエーブレット変換するウエーブレット変換演算手段と、所定信号成分の変換結果に基づいて、所定値以上のピーク間から基本周波数を演算するピッチ演算手段と、所定変動率と基本周波数の変動率とを比較して、変動率の大きさを判定するピッチ変動判定手段とを含むものである。
【0008】
【発明の実施の形態】
実施の形態1.
以下、図面を参照しながら、この発明の実施の形態1について詳細に説明する。図1は、この発明の実施の形態1を示す事故音検出回路の機能ブロック図である。
図1において、10は事故音取込手段、20はピッチ変動音除去手段、30はピッチ変動判定演算手段を示す。100は例えば交差点において検出された交通音の音声信号、110は事故音取込手段10の出力信号、120はピッチ変動判定演算手段30の出力信号、130はピッチ変動音除去手段20の出力信号を示す。
【0009】
事故音取込手段10は、交通音の音声信号100を入力し、事故音と考えられる音声信号を取り込む。ピッチ変動判定演算手段30は、事故音取込手段10からの出力信号110からピッチ変動の大小を判定する。ピッチ変動音除去手段20は、ピッチ変動判定演算手段20の出力信号120に基づいて、事故音取込手段10の出力信号110からピッチ変動の大きい音(ピッチ変動音)の信号成分、例えばサイレン音の信号成分を除去し、急ブレーキ音などの事故音に対応した音声信号を出力する。
【0010】
図2は、この発明の実施の形態1における事故音取込手段10の構成を示す回路図である。
図2において、11は波形成型回路、12はA/D変換器、13は差分演算手段、14は比較回路、15は基準差分値出力回路、16はスイッチを示す。
また、101は波形整形回路の出力信号、102はA/D変換器のディジタル出力信号、103は差分演算手段13から出力される差分信号、104は比較回路14から出力されるトリガー信号を示す。
【0011】
図2を参照しながら、事故取込手段10の動作について説明する。
波形整形回路11は、事故発生時に発生した音声信号100が入力されると、不要な信号成分を除去した出力信号101を生成する。
A/D変換器22は、アナログ信号である出力信号101をディジタル信号102に変換する。
差分演算手段13は、ディジタル信号102を入力し、差分信号103を出力する。差分信号103は一般に信号レベルの変化分を表し、急激なレベル変化があるほど大きな信号レベルを発生する。逆に信号レベルが大きくてもレベル変化が小さい場合にはそのレベルは小さい。従って、例えば事故発生時のように突発的な現象が起こると差分信号103は大きなレベル変化を示すこととなる。
【0012】
比較回路14は、基準差分値出力回路15から出力される基準差分値と、差分信号103とを比較して、差分信号103が基準差分値より大きい場合には信号レベルが急激に変化する事故音が発生したものと判断し、トリガー信号を出力する。
スイッチ16は、トリガー信号によりオンされ、A/D変換器12からのディジタル出力信号102を出力する。
【0013】
図3は、この発明の実施の形態1におけるピッチ変動判定演算手段30の構成を示す回路図である。
図3において、31はウエーブレット変換演算手段、32はピッチ演算手段、33はピッチ変動判定手段を示す。
ウエーブレット変換演算手段31は、事故音取込手段10からの出力信号110を所定時間間隔でウエーブレット変換する。ピッチ演算手段32は、ウエーブレット変換演算手段31における変換結果に基づいてピッチ(基本周波数)f0を演算する。ピッチ変動判定手段33は、ピッチ演算手段32にて所定間隔で演算されたピッチの変動率を演算し、ピッチ変動の大小を判定する。
【0014】
次に、図3とともに、図4から図8を参照しながら、ピッチ変動判定演算手段30の動作について説明する。
【0015】
ウエーブレット変換演算手段31は、A/D変換器12の出力信号102を入力し、所定時間間隔で出力信号102をウエーブレット変換する。
例えば、信号x(t)の2進ウエーブレット変換値DWTは、2進数列で量子化したスケールパラメータa=2およびシフトパラメータbを用いて、式(1)により算出される。
【数1】

Figure 2004126855
なお、式(1)において、g(x)はウエーブレット関数、g(x)はウエーブレット関数g(x)の複素共役である。
図4は、ある時間において、スケールパラメータa=2で、音声信号x(t)をウエーブレット関数g(x)で変換したときの変換結果を示す説明図である。
【0016】
図4において、縦軸はウエーブレット変換値DWTの振幅(amplitude)、横軸は、シフトパラメータ(サンプル数)bを示す。
ウエーブレット変換演算手段31にて、音声信号x(t)をウエーブレット変換すると、音声信号x(t)には、各箇所に振幅のピークが顕著に発生する。そこで、ピッチ演算手段32は、ピークを検出するための閾値A0が設定されており、各ピークの振幅値と閾値A0とを比較して、閾値A0を超えるピークのみを選択する。
続いて、閾値A0を超えるピークの最初と2番目に現れたピークを対象に、ピーク間の周期(ピッチ周期)tp(=サンプル数×サンプル間隔の時間)をそれぞれ算出する。
以下の式(2)において、ピッチ周期tpの逆数を算出することにより、ピッチ周波数f0が算出される。また、上述したピッチ周期の算出において、閾値A0を超えた全てのピークを対象に、各隣り合うピーク間のピッチを算出し、その平均したものをtpとしてもよい。
f0= 1 / tp                 ・・・ (2)
【0017】
このように、ウエーブレット変換演算手段31は、所定間隔でA/D変換器12からの出力信号102をウエーブレット変換していき、ピッチ演算手段32は、ウエーブレット変換演算手段31による変換結果に基づいてピッチ周波数f0の算出を行う。
図5は、ある信号のピッチ周波数の変動を示した説明図であり、図6は、ピッチ周波数の変動率の判定方法を示す説明図である。なお、図5において、縦軸は周波数(kHz)、横軸は時間(sec)を示す。図6において、縦軸はピッチ周波数の変動率、横軸は時間(sec)を示す。
【0018】
ピッチ演算手段32は、ピッチ周波数f0を算出し、図5のように、時間t1におけるピッチ周波数f1、時間t2におけるピッチ周波数f2というように、所定時間におけるピッチ周波数を算出する。
ピッチ変動判定手段33は、以下の式(3)において、時間t1および時間t2間におけるピッチ変動率rpを算出する。
rp = (f2 − f1) / (t2−t1)    ・・・ (3)
【0019】
さらに、ピッチ変動率rpの大小を判定し、算出されたピッチ変動率rpが所定のピッチ周波数Δrの範囲内であれば、ピッチ変動率rpは小さい、すなわち、事故音取込手段10で取り込んだ信号は、急ブレーキ音(事故音)であると判定する。
また、ピッチ変動率rpが所定のピッチ周波数Δrの範囲外であれば、ピッチ変動率rpは大きい、すなわち、事故音取込手段10で取り込んだ信号は、サイレン音であると判定する。
ピッチ変動判定手段33は、ピッチ変動の判定結果をピッチ変動音除去手段20に出力する。
【0020】
図7はサイレン音のピッチ周波数f0を示す説明図であり、図8は急ブレーキ音のピッチ周波数f0を示す説明図である。なお、図7、8において、縦軸は周波数(kHz)、横軸は時間(sec)を示す。
一般にサイレン音は、図7のように、時間の経過とともにピッチ周波数f0が変動しており、急ブレーキ音は、図8のように、時間の経過に関わらずピッチ周波数f0は一定である。
ただし、急ブレーキ音のピッチ周波数が一定と言っても、微少な変化があるので、図6では許容値Δrを設けている。
ピッチ変動判定手段33は、ピッチ周波数f0の変動率を算出することにより、ピッチ周波数f0の変動の大きいサイレン音と、変動のない(小さい)急ブレーキ音との判別が可能となる。
【0021】
ピッチ変動音除去手段20では、ピッチ変動判定演算手段30からの判定結果を入力し、事故音取込手段10で取り込まれた信号において、ピッチ変動が大きい、すなわち、サイレン音であると判定された場合には、事故音取込手段10で取り込まれた信号から、サイレン音と判定された信号成分を除去した信号130を出力する。
また、事故音取込手段10で取り込まれた信号において、ピッチ変動が小さい、すなわち、急ブレーキ音(事故音)であると判定された場合には、その信号成分を除去することなく、例えばこの事故音を記録するためのトリガー信号用の出力信号130として出力する。
【0022】
このように、入力される事故音などの音声信号の中で、急ブレーキ音とサイレン音とを識別するためのピッチ周波数f0を、ウエーブレット変換を用いて算出するので、これまで識別不可能であった急ブレーキ音とサイレン音との識別が可能となり、事故音の検出能力を向上させることができる。
【0023】
【発明の効果】
以上のように、この発明によれば、交通音の音声信号から、信号レベルの変動量が所定レベル以上の所定信号成分を抽出する事故音取込手段と、所定信号成分により基本周波数の変動率の大小を判定するピッチ変動判定演算手段と、変動率が所定変動率よりも大きい場合、所定信号成分を除去し、変動率が所定変動率以下の場合、所定信号成分を事故音として出力するピッチ変動音除去手段とを備え、ピッチ変動判定演算手段は、所定信号成分を所定間隔でウエーブレット変換するウエーブレット変換演算手段と、所定信号成分の変換結果に基づいて、所定値以上のピーク間から基本周波数を演算するピッチ演算手段と、所定変動率と基本周波数の変動率とを比較して、変動率の大きさを判定するピッチ変動判定手段とを含むので、急ブレーキ音とサイレン音との識別が可能となり、事故音の検出能力を向上させることのできる事故音検出回路が得られる効果がある。
【図面の簡単な説明】
【図1】この発明の実施の形態1を示す事故音検出回路の機能ブロック図である。
【図2】この発明の実施の形態1における事故音取込手段の構成を示す回路図である。
【図3】この発明の実施の形態1におけるピッチ変動判定演算手段の構成を示す回路図である。
【図4】この発明の実施の形態1におけるウエーブレット変換による変換結果を示す説明図である。
【図5】この発明の実施の形態1におけるピッチ周波数の変動を示した説明図である。
【図6】この発明の実施の形態1におけるピッチ周波数の変動率の判定方法を示す説明図である。
【図7】この発明の実施の形態1におけるサイレン音のピッチ周波数を示す説明図である。
【図8】この発明の実施の形態1における急ブレーキ音のピッチ周波数を示す説明図である。
【図9】従来の交通事故状況記録装置を示すブロック図である。
【符号の説明】
10 事故音取込手段、20 ピッチ変動音除去手段、30 ピッチ変動判定演算手段、31 ウエーブレット変換演算手段、32 ピッチ演算手段、33 ピッチ変動判定手段、100、101、102 出力信号、103 差分信号、104 トリガー信号、110、120 出力信号、130 トリガー信号、f0、f1、f2 ピッチ周波数、tp ピッチ周波数、rp 変動率。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an accident sound detection circuit for efficiently detecting only an accident sound (such as a sudden brake sound) from traffic sound at an intersection or the like.
[0002]
[Prior art]
In general, many traffic accidents occur at intersections, and there has conventionally been a traffic accident status recording device for recording the status of accidents. The traffic accident status recording apparatus is provided with an accident sound detection circuit for detecting only an accident sound from ambient sounds, and various methods have been proposed.
FIG. 9 is a block diagram showing a conventional traffic accident situation recording apparatus.
9, 1 and 4 are traffic lights, 1B, 1Y, 1R, 4B, 4Y, 4R are signal lights, 2B, 2Y, 2R are signal lines, 3 is a signal light blinking device, 5 is a signal light symbol generator, and 6 is a time letter. A generating unit, 7 is an adding unit, 8 is an imaging device, 9 is an endless recording device, 10 is an ambient sound detecting device, 11 is a collision sound detecting circuit, 12 is a recording control device, and 13 is a magnetic recording / reproducing device.
[0003]
The ambient sound detection device 10 detects a sound around the intersection (traffic sound) and transmits an audio signal corresponding to the traffic sound to the collision sound detection circuit 11. The collision sound detection circuit 11 compares the signal level of the traffic sound audio signal with a threshold value, and determines that it is a vehicle collision sound (accident sound) when the signal level is equal to or higher than the threshold value.
When the accident sound is detected from the traffic sound, the recording control device 12 reproduces the video of the intersection recorded in the endless recording device.
As described above, the conventional accident sound detection circuit detects the accident sound by comparing the signal level of the sound signal of the traffic sound with the threshold value (see, for example, Patent Document 1).
[0004]
[Patent Document 1]
JP-A-4-338900 (page 4, FIG. 2)
[0005]
[Problems to be solved by the invention]
As described above, the conventional accident sound detection circuit determines the accident sound by comparing the signal level of the detected audio signal with a threshold value. There are a lot of loud sounds. Therefore, in the method of comparing the level between the detected audio signal and the threshold value, there is a problem that the siren sound is determined as the accident sound in addition to the actual accident sound such as a sudden brake sound. .
[0006]
The present invention has been made to solve the above problems, and to obtain an accident sound detection circuit capable of distinguishing between a sudden braking sound and a siren sound and improving the detection ability of the accident sound. With the goal.
[0007]
[Means for Solving the Problems]
The accident sound detection circuit according to the present invention includes an accident sound capturing means for extracting a predetermined signal component having a signal level fluctuation amount equal to or higher than a predetermined level from a traffic sound audio signal, and a basic signal fluctuation rate of the basic frequency by the predetermined signal component. Pitch fluctuation determination calculating means for determining the magnitude and pitch fluctuation that removes a predetermined signal component when the fluctuation rate is larger than the predetermined fluctuation rate, and outputs the predetermined signal component as an accident sound when the fluctuation rate is equal to or lower than the predetermined fluctuation rate And a sound fluctuation removing means, wherein the pitch fluctuation determining calculating means is based on a wavelet conversion calculating means for wavelet converting predetermined signal components at predetermined intervals, and based on a conversion result of the predetermined signal components from a peak between predetermined peaks or more. Pitch calculation means for calculating the frequency and pitch fluctuation determination means for comparing the predetermined fluctuation rate and the fluctuation rate of the fundamental frequency to determine the magnitude of the fluctuation rate are included.
[0008]
DETAILED DESCRIPTION OF THE INVENTION
Embodiment 1 FIG.
Hereinafter, the first embodiment of the present invention will be described in detail with reference to the drawings. 1 is a functional block diagram of an accident sound detection circuit showing Embodiment 1 of the present invention.
In FIG. 1, 10 is an accident sound capturing means, 20 is a pitch fluctuation sound removing means, and 30 is a pitch fluctuation determination calculating means. 100 is an audio signal of traffic sound detected at, for example, an intersection, 110 is an output signal of the accident sound capturing means 10, 120 is an output signal of the pitch fluctuation determination calculating means 30, and 130 is an output signal of the pitch fluctuation sound removing means 20. Show.
[0009]
The accident sound capturing means 10 receives the traffic sound audio signal 100 and captures an audio signal considered to be an accident sound. The pitch fluctuation determination calculating means 30 determines the magnitude of the pitch fluctuation from the output signal 110 from the accident sound capturing means 10. Based on the output signal 120 of the pitch fluctuation determination calculating means 20, the pitch fluctuation sound removing means 20 is a signal component of a sound having a large pitch fluctuation (pitch fluctuation sound) from the output signal 110 of the accident sound capturing means 10, for example, a siren sound. The audio signal corresponding to the accident sound such as a sudden brake sound is output.
[0010]
FIG. 2 is a circuit diagram showing a configuration of accident sound capturing means 10 according to Embodiment 1 of the present invention.
In FIG. 2, 11 is a waveform shaping circuit, 12 is an A / D converter, 13 is a difference calculation means, 14 is a comparison circuit, 15 is a reference difference value output circuit, and 16 is a switch.
Reference numeral 101 denotes an output signal of the waveform shaping circuit, 102 denotes a digital output signal of the A / D converter, 103 denotes a difference signal output from the difference calculation means 13, and 104 denotes a trigger signal output from the comparison circuit 14.
[0011]
The operation of the accident taking means 10 will be described with reference to FIG.
When the audio signal 100 generated at the time of the accident is input, the waveform shaping circuit 11 generates an output signal 101 from which unnecessary signal components are removed.
The A / D converter 22 converts the output signal 101 that is an analog signal into a digital signal 102.
The difference calculation means 13 receives the digital signal 102 and outputs a difference signal 103. The difference signal 103 generally represents a change in the signal level, and a larger signal level is generated as there is a sudden level change. Conversely, if the level change is small even if the signal level is large, the level is small. Therefore, for example, when a sudden phenomenon occurs such as when an accident occurs, the difference signal 103 shows a large level change.
[0012]
The comparison circuit 14 compares the reference difference value output from the reference difference value output circuit 15 with the difference signal 103, and when the difference signal 103 is larger than the reference difference value, the accident sound whose signal level changes rapidly. The trigger signal is output.
The switch 16 is turned on by the trigger signal and outputs the digital output signal 102 from the A / D converter 12.
[0013]
FIG. 3 is a circuit diagram showing a configuration of pitch variation determination calculation means 30 according to the first embodiment of the present invention.
In FIG. 3, 31 is a wavelet conversion calculating means, 32 is a pitch calculating means, and 33 is a pitch fluctuation determining means.
The wavelet conversion calculation means 31 performs wavelet conversion on the output signal 110 from the accident sound capturing means 10 at predetermined time intervals. The pitch calculation means 32 calculates the pitch (basic frequency) f0 based on the conversion result in the wavelet conversion calculation means 31. The pitch fluctuation determination means 33 calculates the pitch fluctuation rate calculated at predetermined intervals by the pitch calculation means 32 and determines the magnitude of the pitch fluctuation.
[0014]
Next, the operation of the pitch variation determination calculation means 30 will be described with reference to FIGS. 4 to 8 together with FIG.
[0015]
The wavelet conversion calculation means 31 receives the output signal 102 of the A / D converter 12 and wavelet converts the output signal 102 at predetermined time intervals.
For example, the binary wavelet transform value D y WT of the signal x (t) is calculated by equation (1) using the scale parameter a = 2 j and the shift parameter b quantized with a binary number sequence.
[Expression 1]
Figure 2004126855
In Equation (1), g (x) is a wavelet function, and g * (x) is a complex conjugate of the wavelet function g (x).
FIG. 4 is an explanatory diagram showing a conversion result when the audio signal x (t) is converted by the wavelet function g (x) with the scale parameter a = 2 4 at a certain time.
[0016]
In FIG. 4, the vertical axis represents the amplitude (amplitude) of the wavelet transform value D y WT, and the horizontal axis represents the shift parameter (number of samples) b.
When the wavelet transform calculation means 31 performs wavelet transform on the audio signal x (t), the audio signal x (t) has a significant amplitude peak at each location. Therefore, the pitch calculation means 32 is set with a threshold value A0 for detecting a peak, and compares only the peak value exceeding the threshold value A0 by comparing the amplitude value of each peak with the threshold value A0.
Subsequently, with respect to the first and second peaks that exceed the threshold A0, the period between the peaks (pitch period) tp (= number of samples × sample interval time) is calculated.
In the following equation (2), the pitch frequency f0 is calculated by calculating the reciprocal of the pitch period tp. Further, in the above-described calculation of the pitch period, the pitch between adjacent peaks may be calculated for all peaks exceeding the threshold A0, and the average may be used as tp.
f0 = 1 / tp (2)
[0017]
As described above, the wavelet conversion calculation unit 31 performs wavelet conversion on the output signal 102 from the A / D converter 12 at a predetermined interval, and the pitch calculation unit 32 converts the conversion result obtained by the wavelet conversion calculation unit 31 into the conversion result. Based on this, the pitch frequency f0 is calculated.
FIG. 5 is an explanatory diagram showing the variation of the pitch frequency of a certain signal, and FIG. 6 is an explanatory diagram showing a method of determining the variation rate of the pitch frequency. In FIG. 5, the vertical axis represents frequency (kHz) and the horizontal axis represents time (sec). In FIG. 6, the vertical axis represents the variation rate of the pitch frequency, and the horizontal axis represents time (sec).
[0018]
The pitch calculation means 32 calculates the pitch frequency f0, and calculates the pitch frequency at a predetermined time, such as the pitch frequency f1 at time t1 and the pitch frequency f2 at time t2, as shown in FIG.
The pitch variation determination means 33 calculates a pitch variation rate rp between time t1 and time t2 in the following equation (3).
rp = (f2-f1) / (t2-t1) (3)
[0019]
Further, the magnitude of the pitch fluctuation rate rp is determined, and if the calculated pitch fluctuation rate rp is within the predetermined pitch frequency Δr, the pitch fluctuation rate rp is small, that is, captured by the accident sound capturing means 10. It is determined that the signal is a sudden brake sound (accident sound).
If the pitch fluctuation rate rp is outside the range of the predetermined pitch frequency Δr, the pitch fluctuation rate rp is large, that is, it is determined that the signal captured by the accident sound capturing means 10 is a siren sound.
The pitch fluctuation determining unit 33 outputs the pitch fluctuation determination result to the pitch fluctuation sound removing unit 20.
[0020]
FIG. 7 is an explanatory diagram showing the pitch frequency f0 of the siren sound, and FIG. 8 is an explanatory diagram showing the pitch frequency f0 of the sudden brake sound. 7 and 8, the vertical axis represents frequency (kHz) and the horizontal axis represents time (sec).
In general, the pitch frequency f0 of the siren sound fluctuates with time as shown in FIG. 7, and the pitch frequency f0 of the sudden brake sound is constant regardless of the passage of time as shown in FIG.
However, even if the pitch frequency of the sudden braking sound is constant, there is a slight change, so the allowable value Δr is provided in FIG.
The pitch variation determination means 33 can distinguish between a siren sound having a large variation in the pitch frequency f0 and a sudden brake sound having no variation (small) by calculating the variation rate of the pitch frequency f0.
[0021]
In the pitch fluctuation sound removing means 20, the judgment result from the pitch fluctuation judgment calculating means 30 is inputted, and the signal taken in by the accident sound taking means 10 is judged to have a large pitch fluctuation, that is, a siren sound. In this case, a signal 130 obtained by removing a signal component determined to be a siren sound from the signal captured by the accident sound capturing means 10 is output.
Further, in the signal captured by the accident sound capturing means 10, when it is determined that the pitch fluctuation is small, that is, a sudden brake sound (accident sound), for example, this signal component is removed without removing the signal component. Output as an output signal 130 for a trigger signal for recording the accident sound.
[0022]
As described above, the pitch frequency f0 for discriminating between the sudden braking sound and the siren sound in the input audio signal such as an accident sound is calculated using the wavelet transform, so that it cannot be identified so far. It is possible to distinguish between the sudden braking sound and the siren sound, and it is possible to improve the detection ability of the accident sound.
[0023]
【The invention's effect】
As described above, according to the present invention, accident sound capturing means for extracting a predetermined signal component having a signal level variation amount equal to or higher than a predetermined level from a traffic sound audio signal, and a fundamental frequency variation rate by the predetermined signal component. Pitch fluctuation determination calculation means for determining the magnitude of the pitch, and a pitch for removing the predetermined signal component when the fluctuation rate is greater than the predetermined fluctuation rate and outputting the predetermined signal component as an accident sound when the fluctuation rate is equal to or less than the predetermined fluctuation rate And a fluctuation sound removing means, wherein the pitch fluctuation determining calculating means is a wavelet conversion calculating means for wavelet converting predetermined signal components at predetermined intervals, and based on a conversion result of the predetermined signal components, between pitches greater than a predetermined value. Since the pitch calculation means for calculating the fundamental frequency and the pitch fluctuation determination means for comparing the predetermined fluctuation rate and the fluctuation rate of the fundamental frequency to determine the magnitude of the fluctuation rate, It is possible to distinguish between sound and siren sound, the accident sound detection circuit capable of improving the detection capability of an accident sound is the effect obtained.
[Brief description of the drawings]
FIG. 1 is a functional block diagram of an accident sound detection circuit showing a first embodiment of the present invention.
FIG. 2 is a circuit diagram showing a configuration of an accident sound capturing means in Embodiment 1 of the present invention.
FIG. 3 is a circuit diagram showing a configuration of pitch fluctuation determination calculation means according to the first embodiment of the present invention.
FIG. 4 is an explanatory diagram showing a conversion result by wavelet conversion according to Embodiment 1 of the present invention;
FIG. 5 is an explanatory diagram showing variation in pitch frequency according to the first embodiment of the present invention.
FIG. 6 is an explanatory diagram illustrating a pitch frequency variation rate determination method according to Embodiment 1 of the present invention;
FIG. 7 is an explanatory diagram showing the pitch frequency of the siren sound according to the first embodiment of the present invention.
FIG. 8 is an explanatory diagram showing a pitch frequency of a sudden brake sound according to the first embodiment of the present invention.
FIG. 9 is a block diagram showing a conventional traffic accident situation recording apparatus.
[Explanation of symbols]
10 Accident sound capture means, 20 Pitch fluctuation sound removal means, 30 Pitch fluctuation determination calculation means, 31 Wavelet conversion calculation means, 32 Pitch calculation means, 33 Pitch fluctuation determination means, 100, 101, 102 Output signal, 103 Difference signal 104 trigger signal, 110, 120 output signal, 130 trigger signal, f0, f1, f2 pitch frequency, tp pitch frequency, rp variation rate.

Claims (1)

交通音の音声信号の中から事故音を検出するための事故音検出回路であって、
前記音声信号から、信号レベルの変動量が所定レベル以上を示す所定信号成分を抽出する事故音取込手段と、
前記所定信号成分に基づいて、基本周波数の変動率の大小を判定するピッチ変動判定演算手段と、
前記変動率が所定変動率よりも大きい場合には、前記音声信号から前記所定信号成分を除去するとともに、前記変動率が前記所定変動率以下の場合には、前記所定信号成分を前記事故音と判定して出力するピッチ変動音除去手段とを備え、
前記ピッチ変動判定演算手段は、
前記所定信号成分を所定間隔でウエーブレット変換するウエーブレット変換演算手段と、
前記所定信号成分のウエーブレット変換結果に基づいて、所定値以上のピークを選択し、前記ピーク間から前記基本周波数を演算するピッチ演算手段と、
前記基本周波数の変動率を算出し、前記所定変動率と前記変動率とを比較して、前記変動率の大きさを判定するピッチ変動判定手段と
を含むことを特徴とする事故音検出回路。
An accident sound detection circuit for detecting an accident sound from an audio signal of traffic sound,
Accident sound capturing means for extracting a predetermined signal component having a signal level variation amount equal to or higher than a predetermined level from the audio signal;
Pitch fluctuation determination calculating means for determining the magnitude of the fluctuation rate of the fundamental frequency based on the predetermined signal component;
When the fluctuation rate is greater than a predetermined fluctuation rate, the predetermined signal component is removed from the audio signal, and when the fluctuation rate is equal to or less than the predetermined fluctuation rate, the predetermined signal component is defined as the accident sound. And a pitch fluctuation sound removing means for determining and outputting,
The pitch fluctuation determination calculating means includes:
Wavelet transform computing means for wavelet transforming the predetermined signal component at predetermined intervals;
Based on the wavelet conversion result of the predetermined signal component, a pitch calculating means for selecting a peak greater than a predetermined value and calculating the fundamental frequency between the peaks;
An accident sound detection circuit comprising: pitch fluctuation determining means for calculating a fluctuation rate of the fundamental frequency and comparing the predetermined fluctuation rate with the fluctuation rate to determine the magnitude of the fluctuation rate.
JP2002288517A 2002-10-01 2002-10-01 Accident sound detection circuit Expired - Lifetime JP3895657B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2002288517A JP3895657B2 (en) 2002-10-01 2002-10-01 Accident sound detection circuit

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2002288517A JP3895657B2 (en) 2002-10-01 2002-10-01 Accident sound detection circuit

Publications (2)

Publication Number Publication Date
JP2004126855A true JP2004126855A (en) 2004-04-22
JP3895657B2 JP3895657B2 (en) 2007-03-22

Family

ID=32280993

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2002288517A Expired - Lifetime JP3895657B2 (en) 2002-10-01 2002-10-01 Accident sound detection circuit

Country Status (1)

Country Link
JP (1) JP3895657B2 (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007080764A1 (en) * 2006-01-12 2007-07-19 Matsushita Electric Industrial Co., Ltd. Object sound analysis device, object sound analysis method, and object sound analysis program
CN110176248A (en) * 2019-05-23 2019-08-27 广西交通科学研究院有限公司 Road sound identification method, system, computer equipment and readable storage medium storing program for executing
CN115223370A (en) * 2022-08-31 2022-10-21 四川九通智路科技有限公司 Traffic accident detection method and system

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007080764A1 (en) * 2006-01-12 2007-07-19 Matsushita Electric Industrial Co., Ltd. Object sound analysis device, object sound analysis method, and object sound analysis program
US8223978B2 (en) 2006-01-12 2012-07-17 Panasonic Corporation Target sound analysis apparatus, target sound analysis method and target sound analysis program
CN110176248A (en) * 2019-05-23 2019-08-27 广西交通科学研究院有限公司 Road sound identification method, system, computer equipment and readable storage medium storing program for executing
CN115223370A (en) * 2022-08-31 2022-10-21 四川九通智路科技有限公司 Traffic accident detection method and system
CN115223370B (en) * 2022-08-31 2023-01-17 四川九通智路科技有限公司 Traffic accident detection method and system

Also Published As

Publication number Publication date
JP3895657B2 (en) 2007-03-22

Similar Documents

Publication Publication Date Title
JP4242422B2 (en) Sudden event recording and analysis system
US5990801A (en) Accident sound detection circuit
JP2012027186A (en) Sound signal processing apparatus, sound signal processing method and program
WO2004114278A1 (en) System and method for spectrogram analysis of an audio signal
JP2007280407A (en) Traffic terminal device and accident detection system
KR960005741B1 (en) Voice signal coding system
US6087960A (en) Accident sound detection circuit
JP3895657B2 (en) Accident sound detection circuit
JP2003202260A (en) Sound source identifying device, sudden event detecting device, and device for automatically recording sudden event
US20060012831A1 (en) Electronic watermarking method and storage medium for storing electronic watermarking program
JP2005284308A (en) Audio information classifying device
JP4392805B2 (en) Audio information classification device
JP3845034B2 (en) Accident sound detection device
DE60325736D1 (en) Method and apparatus for noise reduction in a sound signal
KR100659884B1 (en) Method on automatic detection of vibrato in music
JP3895621B2 (en) Accident sound detection device
JP2950771B2 (en) Accident sound detection circuit and accident sound detection device
JP4652165B2 (en) Train control receiver
JP3810332B2 (en) Accident sound detection device
JP3015298B2 (en) Accident sound detection circuit and accident sound detection device
JP5177657B2 (en) Acoustic characteristic control device
JP3190231B2 (en) Apparatus and method for extracting pitch period of voiced sound signal
SU1781701A1 (en) Method of separation of speech and nonstationary noise signals
CN117520853A (en) Pedestrian subjective intention judging method and system based on multidimensional information fusion
JP4497485B2 (en) Audio information classification device

Legal Events

Date Code Title Description
A621 Written request for application examination

Free format text: JAPANESE INTERMEDIATE CODE: A621

Effective date: 20050523

TRDD Decision of grant or rejection written
A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

Effective date: 20061114

A61 First payment of annual fees (during grant procedure)

Free format text: JAPANESE INTERMEDIATE CODE: A61

Effective date: 20061214

R150 Certificate of patent or registration of utility model

Free format text: JAPANESE INTERMEDIATE CODE: R150

Ref document number: 3895657

Country of ref document: JP

Free format text: JAPANESE INTERMEDIATE CODE: R150

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

Free format text: PAYMENT UNTIL: 20091222

Year of fee payment: 3

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

Free format text: PAYMENT UNTIL: 20101222

Year of fee payment: 4

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

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

Free format text: PAYMENT UNTIL: 20101222

Year of fee payment: 4

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

Free format text: PAYMENT UNTIL: 20111222

Year of fee payment: 5

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

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

Free format text: PAYMENT UNTIL: 20121222

Year of fee payment: 6

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

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

Free format text: PAYMENT UNTIL: 20131222

Year of fee payment: 7

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

EXPY Cancellation because of completion of term