WO2009140792A1 - 降低音频干扰的方法及其装置 - Google Patents

降低音频干扰的方法及其装置 Download PDF

Info

Publication number
WO2009140792A1
WO2009140792A1 PCT/CN2008/000980 CN2008000980W WO2009140792A1 WO 2009140792 A1 WO2009140792 A1 WO 2009140792A1 CN 2008000980 W CN2008000980 W CN 2008000980W WO 2009140792 A1 WO2009140792 A1 WO 2009140792A1
Authority
WO
WIPO (PCT)
Prior art keywords
pulse
audio
noise
template
steps
Prior art date
Application number
PCT/CN2008/000980
Other languages
English (en)
French (fr)
Inventor
林翰
Original Assignee
Lin Han
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 Lin Han filed Critical Lin Han
Priority to PCT/CN2008/000980 priority Critical patent/WO2009140792A1/zh
Priority to CN2008801293036A priority patent/CN102037664A/zh
Publication of WO2009140792A1 publication Critical patent/WO2009140792A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04BTRANSMISSION
    • H04B1/00Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission
    • HELECTRICITY
    • H03ELECTRONIC CIRCUITRY
    • H03GCONTROL OF AMPLIFICATION
    • H03G3/00Gain control in amplifiers or frequency changers
    • H03G3/20Automatic control
    • H03G3/30Automatic control in amplifiers having semiconductor devices
    • H03G3/34Muting amplifier when no signal is present or when only weak signals are present, or caused by the presence of noise signals, e.g. squelch systems

Definitions

  • the present invention relates to a method and apparatus for reducing interference, and more particularly to a method and apparatus for reducing audio interference. Background technique
  • GSM buzz global mobile communication system abnormal tone
  • the present invention provides a method and apparatus for reducing interference of noise pulses to an audio device, such as interference sources such as mobile phones or cellular phones.
  • the concept is mainly to first find a pattern pulse of a noise pulse caused by a mobile phone or a cellular phone, and then pulse the template from the audio device that is affected by the noise pulse and contains the noise pulse waveform.
  • the audio is subtracted, and then the processed audio is interpolated to positive, such as auto-regressive interpolation, and then the noise is applied to the corrected audio, such as the spectrum.
  • noise suppression techniques such as techniques, the above steps, when implemented, will be classified as short bursts or long bursts depending on the duration of the noise pulses, with different possible combinations.
  • a method of reducing audio interference is provided that is subject to one or a series of similar noise pulses, the noise pulse comprising a central pulse and an extended pulse characterized by: generating the same plate pulse Representing the noise pulse; and performing the following steps (A), (B), (C), or a combination thereof to suppress the noise pulse: (A) subtracting the template pulse or the template from the position of the noise pulse in the audio Generating a first processed audio in the pulse; (B) interpolating the first processed audio or the second processed audio generated by the audio at least across the central pulse of the noise pulse; or (C) The second processed audio or the audio is processed by a noise suppression technique.
  • the method for reducing audio interference provided by the present invention is characterized in that the sound is The frequency is divided into a plurality of intervals such as frames, and a step of suppressing the noise pulse is implemented in each frame.
  • the method of reducing audio interference provided by the present invention is characterized in that one or more consecutive noise pulses are classified into long bursts or short bursts according to their duration.
  • the method for reducing audio interference provided by the present invention is characterized in that: if the noise pulse is the short burst, then steps (A), (B) or a combination thereof is implemented.
  • the method for reducing audio interference provided by the present invention is characterized in that: if the noise pulse is the long burst, step (A), (B), (C) or a combination thereof is implemented. .
  • the method for reducing audio interference provided by the present invention, wherein the generating step further comprises the steps of: confirming a position of the noise pulse in the audio and an amplitude of the noise pulse, thereby calculating according to the amplitude
  • the scaling of the template pulse is then subtracted from the template pulse or the central pulse of the template pulse at the corresponding location of the audio.
  • the method for reducing audio interference provided by the present invention is characterized in that the generating step further comprises the step of: estimating a plurality of features of the noise pulse to thereby generate the template pulse.
  • the method of reducing audio interference provided by the present invention is characterized in that the interpolation utilizes an autoregressive mode, a currently commercially available recovery technique or a combination thereof.
  • the method of reducing audio interference provided by the present invention is characterized in that the noise suppression technique is a spectrum subtraction technique, a currently commercially available noise suppression technique or a combination thereof.
  • the method for reducing audio interference provided by the present invention is characterized in that the noise pulse is generated by a mobile phone or a cellular phone.
  • the method for reducing audio interference provided by the present invention is characterized in that the audio is audio emitted by an audio device, and the audio device is a microphone, an audio, a telephone, a recorder, and a Portable audio player or a hearing aid.
  • an apparatus for reducing audio interference comprising: an input for receiving an audio; a memory for generating a same board pulse for storing a noise pulse, wherein the The template pulse represents one of a series of similar noise pulses, the noise pulse including a central pulse and an extended pulse; and a noise suppressor coupled to the input for performing the following steps (A;), (B), (C) or a combination thereof to suppress the noise pulse: (A) subtracting the audio from the audio Generating a first processed audio by the template pulse; (B) interpolating the first processed audio or the audio to generate a second processed audio at least across the central pulse of the noise pulse; ⁇ / or (C) with a noise
  • the suppression technique processes the second processed audio or the audio.
  • the apparatus for reducing audio interference provided by the present invention is characterized by further comprising a detector for confirming the occurrence of a noise pulse.
  • the apparatus for reducing audio interference provided by the present invention is characterized in that the noise suppressor further comprises an amplitude determining means for determining the amplitude of the pulse.
  • the apparatus for reducing audio interference provided by the present invention is characterized in that the noise suppressor further comprises a pulse position detector for detecting a position of the noise pulse, and the pulse position detector Also included is an interactive correlation detector.
  • the present invention minimizes the effect of the noise pulse by decomposing the noise pulse into a central pulse and an extended pulse, wherein the central pulse is an electromagnetic radiation product from the cellular phone or other source of interference,
  • the extension pulse is derived from the characteristics of the audio device, and in particular the capacitance characteristics of the audio device.
  • the audio circuit characteristics of the audio device are not taken into account when the central pulse is sampled and the sampled central pulse is subtracted.
  • Figure 1 is a schematic illustration of a noise pulse typically generated by an interference source such as a mobile phone or a cellular telephone;
  • FIG. 1 is a device for suppressing noise interference from audio in a reduced audio interference
  • FIG. 3 is a flowchart of an implementation of a method for reducing audio interference according to the present invention
  • FIG. 4 is a noise pulse classified as a short burst.
  • Figure 5 is a schematic diagram of noise pulses classified as long bursts
  • Figure 6 is a schematic diagram of the spectrum of a noise pulse. Preferred embodiment of the invention
  • FIG. 1 is a schematic diagram of a typical noise pulse generated by an interference source such as a mobile phone or a cellular phone.
  • the noise pulse ⁇ shown in FIG. 1 includes two main parts, a central pulse ⁇ and a The pulse ⁇ 2 is extended.
  • the condition that produces the noise pulse shown in Figure 1 is when an audio device, such as a microphone, stereo, telephone, recorder, portable audio player, or hearing aid is operating, that is, receiving audio or playing audio.
  • an audio device such as a microphone, stereo, telephone, recorder, portable audio player, or hearing aid is operating, that is, receiving audio or playing audio.
  • communication or information transmission is also taking place.
  • a normal or smooth normal audio in the audio device will be received by the mobile phone or the cellular phone.
  • the resulting pulse interferes with the beginning of the original normal audio with the noise pulse as shown in Figure 1, including the central pulse ⁇ and the extended pulse ⁇ 2 .
  • the noise pulse ⁇ whose central pulse ⁇ is excited by electromagnetic waves generated by a mobile phone or a cellular phone during communication or information transmission, and the extension pulse ⁇ 2 is related to the capacitance characteristics of the audio device. . Therefore, the characteristics of the noise pulse ⁇ will be related to the electromagnetic radiation intensity of the cellular phone and the characteristics of the audio circuit of the audio device.
  • the length of the audio pulse transmitted by the mobile phone or the cellular phone varies with the specifications of the phone, generally Said to be equal to 25 samples for a sampling frequency of 44 kHz, the amplitude of the pulse is especially proportional to the distance between the cellular phone and the audio device.
  • Figure 2 is a device for reducing audio interference from noise suppression in audio.
  • an input 10 of the apparatus 20 for reducing audio interference is coupled to an audio circuit 1, which is an audio circuit disposed in various audio devices.
  • the audio device or audio circuit 1 is subject to interference caused by electromagnetic pulses surrounding the mobile phone or the cellular phone.
  • the output of the audio circuit 1 is provided to an analog to digital converter 2, if the audio is already In digital form, for example from an optical disc, the analog to digital converter 2 will be omitted and the output of the audio circuit 1 will be provided directly to a first in first out (FIFO) buffer 3; the analog to digital converter 2
  • the output of the first in first out (FIFO) buffer 3 is connected to a digital signal processor (Digital Signal Processor) 4; a hardware electromagnetic pulse detection
  • the device 5 has an output connected to the digital signal processor 4.
  • the analog to digital converter 2 also converts an output to a memory 6, which memory 6 can also be accessed by the digital signal processor 4.
  • One of the digital signal processors 4 The output is also provided to a digital to analog converter 7, which is coupled to an output 11, which will be combined with some form of transducer such as a speaker or earphone.
  • a digital to analog converter 7 which is coupled to an output 11, which will be combined with some form of transducer such as a speaker or earphone.
  • the combination of the analog-to-digital converter 2, the first-in-first-out (FIFO) buffer 3, the digital signal processor 4, the digital-to-analog converter 7, and the like disclosed in FIG. 2 can also be regarded as a filter 22.
  • a noise suppressor 24 that removes the noise pulses present in input 10 such that the audio of output 11 is audio that is noise free.
  • the digital signal processor 4 also serves as an amplitude determining device for determining the amplitude of the pulse, and as a pulse position detector for detecting the position of the noise pulse.
  • FIG. 3 is a flowchart of an implementation of a method for reducing audio interference according to the present invention.
  • the method for reducing audio interference disclosed in FIG. 3, comprising: Step 30: dividing audio into a plurality of frames; Step 32, generating the same plate pulse to represent the noise pulse; Step 34, the noise pulse Classified as a short burst or a long burst; Step 36, if the noise pulse is the long burst, perform steps (A), (B), (C) or a combination thereof; Step 38, if the noise pulse For the short burst, step (A), (B) or a combination thereof is implemented; step (A) is to generate a first processed audio by subtracting the template pulse from the audio; and step (B) spans at least the noise The central processed pulse or the audio is interpolated to generate a second processed audio; the step (C) processing the second processed audio or the audio in a noise suppression technique.
  • the steps indicated by dashed lines in FIG. 3, such as steps 30, 34 and 38 are
  • an audio is first obtained from an audio device in operation, and the audio device is interfered by a pulse of a mobile phone or a cellular phone in use or in communication, so that the audio contains a noise pulse, and the noise pulse Divided into two parts, the central pulse and the extended pulse, wherein the central pulse is excited by the electromagnetic wave generated by the mobile phone or the cellular phone during communication or information transmission, and the extension pulse is combined with the capacitance characteristic of the audio device. related.
  • the audio signal must first be divided into several frames, each frame being approximately 40/wsec in length. If the above method is not to be implemented in a frame-based manner, the above described method can be ignored. The above is the step 30 shown in Fig. 3. Hereinafter, the above method will be described by determining a frame-based method.
  • the same plate pulse representing a typical pulse generated by a mobile phone or a cellular phone is constructed, and the device shown in Fig. 2 is set to have a zero-valued audio, which is only passed through the audio circuit 1 and supplied to the digital signal.
  • the signal of the processor 4 is a cellular telephone pulse, which represents those noise pulses that interfere with normal audio in normal operation, and calculates the amplitude, period and the like of the pulse, thereby generating a template pulse and storing it in the memory 6. .
  • the scaling of the sample pulse is calculated according to the amplitude, and the template pulse can be scaled up or down to match the noise pulse, and the length of the central pulse and the extended pulse of the noise pulse is A cellular phone of known size is constant with a known specification of audio device, but the intensity (amplitude) of the noise pulse will be related to the distance of the mobile phone or cellular phone from the audio device.
  • the digital signal processor 4 is programmed to perform the methods described herein using the template pulses stored in the memory 6, and simultaneously generate the template pulses as described and update them as needed.
  • the noise pulse measured when it is blank is used as a template pulse
  • the noise pulse obtained by using the previous frame is used as a template pulse.
  • the measured noise pulse is obtained as a template pulse or the like
  • the sample pulse is obtained before the audio step is received. The above is the step 32 shown in FIG.
  • the noise pulse is searched for in the frame and the start of each noise pulse is identified. If more than one noise pulse is found, the frequency of occurrence of the noise pulse in the frame is determined and the noise pulse in the frame is classified as a long burst or a short burst, which usually lasts for more than 30
  • the sec noise pulses are classified as long bursts, and noise pulses that do not last longer than 30/ «sec are classified as short bursts.
  • Figure 4 and Figure 5 for a schematic diagram of noise pulses classified as short bursts and a noise pulse classified as long bursts.
  • step (A) and (B) perform the combined steps of steps (A), (B) or steps (A) and (B) (steps (A) and (B) will be detailed later. That is, steps (A) and (B) are sequentially performed, steps (B) and (A) are sequentially performed, only step (A) is performed, and only step (B) or the like is performed. Usually only one of the steps of step (A) or (B) can be effectively suppressed to suppress noise pulses classified as short bursts, especially when When the extension pulse of the noise pulse is quite obvious, the implementation of step (B) can be omitted.
  • step (A), (B), and (C) will be described in detail later, ie, it is optional to carry out only one of the steps (A), (B) or (C) or two of them, and in the implementation, the steps The order of the order can be alternated according to the actual situation. Another common situation is: If the noise pulse in this frame has been classified as a long burst, it is recommended that step (C) be implemented after the short burst-based step is implemented, or step (C) should be implemented first. , then implement the steps based on short bursts.
  • the present invention uses the following three steps (A), (B), (C) or a combination of (A), (B) and (C) to suppress noise pulses carried in the audio.
  • step (A) in which the mode pulse is subtracted from the position of each noise pulse in the audio to generate a first processed audio.
  • step (A) the noise pulse caused by the cellular phone contained in the audio is removed by using the equation described below:
  • x noise is an audio signal containing a noise pulse caused by the cellular phone
  • the scale factor b can be used to scale up or down the sample pulse and then subtract it from the corresponding position of the audio.
  • the central pulse of the sample pulse or the plate pulse is removed.
  • the original signal x 0 al can be preferably restored.
  • This original signal; C. responsible. g , . fl / can also be temporarily regarded as the first processing audio.
  • the above is the step (A;) shown in Figure 3.
  • step (A) it is necessary to know the exact position of the noise pulse in the audio, and then subtract the center pulse of the scaled sample pulse or the template pulse at this position, usually there are many detections.
  • the method can be used to determine the exact position of the noise pulse, such as: 1) hardware electromagnetic wave detector, 2) maximum threshold detection of signal amplitude, 3) maximum threshold detection of signal slope, 4) Inter-related/conforming filter detectors, 5) Bayesian ladder detectors, and 6) autoregressive detectors.
  • hardware electromagnetic wave detectors may be the most convenient and most effective solution.
  • Software implementations that use signal amplitude threshold detection or signal slope maximum threshold detection are also simple, efficient, and effective solutions.
  • step (B) is illustrated to interpolate the first processed audio or the audio at least across the central pulse of the noise pulse to produce a second processed audio.
  • ⁇ ⁇ can therefore use ⁇ a simple interpolation mode to assume that the information between these 25 samples is lost and interpolated.
  • the second processed audio obtained by interpolation obtained by interpolation.
  • step (B) is not limited to the AR mode, and various other commercially available interpolation techniques for recovering audio can be applied to the step (B), simulating the central pulse. The lost audio has occurred.
  • step (C) is illustrated to process the second processed audio or the audio in a noise suppression technique.
  • the noise suppression techniques indicated in step (C) are current and future commercially available noise suppression techniques such as: frequency subtraction techniques, power spectrum subtraction techniques, and others. And future groups of commercially available noise suppression technologies Hehe. Please refer to Figure 6.
  • the spectrum or power spectrum in each frame can be established by using an external detector directly to estimate the spectrum, using the historical experience to establish the spectrum, before use.
  • the spectrum of one time, the frequency spectrum measured when the audio in the audio device is blank is used as the template frequency consultation, and the spectrum obtained by using the previous frame is used.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Noise Elimination (AREA)
  • Mobile Radio Communication Systems (AREA)

Description

降低音频干扰的方法及其装置
技术领域
本发明是关于一种降低千扰的方法及装置, 特别是关于一种降低音频千 扰的方法及装置。 背景技术
随着通信科技的快速发展, 各种移动电话 (mobile phone)或蜂巢式电话 (cell phone)早已普及, 其所采用的各类传输通信协议所使用的无线传输通信 协议在通信的过程中产生大量的大振幅电磁脉冲, 在注册网路、 接收、 传输 过程以及交递步骤期间, 进行控制或传输数据信号, 一旦这些移动电话中的 语音信号进行传输, 音频脉沖将由音频放大器和线上电路所接收, 从而引起 断音形式的可感知音响扭曲产生,通常称作为全球移动通信系统异常音 (GSM buzz), 并具有相对于全球移动通信系统信号而言约 200赫兹附近的频率。
许多音频设备如汽车立体音响装置、 电话、 录音机、 可携式音频播放器 以及听力辅助器等都会受到这些正在通信中的移动电话或蜂巢式电话的音频 脉沖干扰, 当使用蜂巢式电话的用户数量持续增加时, 这个问题就会变得更 加明显, 特别是对于顶级消费端音频产品以及听力辅助器而言, 音响的清晰 度更是产品品质的重点。
虽然降低蜂巢式电话传输功率, 改变该传输通信协议以及遮蔽该音频电 路可以减轻此问题, 但是这些解决方式无法在不改变现有硬件设计的情况下 实现, 且通常非常困难和昂贵。 此外, 以现今的通信科技而言, 设计一无干 扰音频电路, 使其能够完全免除利用多种通信协议传输电磁脉冲所造成的噪 声干扰几乎是不可能的。 移动电话或蜂巢式电话对这些音频的千扰几乎无处 - 不在, 干扰常常发生在音频传输路径的任何相位之中, 一旦这些音频设备的 音频波形被干扰时, 并不存在即滤除该噪声又不会造成该原始音频的扭曲的 解决方式。
当然, 一种更简单的解决方式是不要在这些音频设备周围使用移动电话 或蜂巢式电话, 但这并不合常理。 举例而言, 如果在汽车中使用蜂巢式电话 时, 其将足够靠近该汽车音频设备而引起干扰, 而如果在汽车内部使用蜂巢 式电话并同时移动时, 其将必然千扰该汽车音频设备。 此外, 如杲一个有听 觉障碍的人, 当其使用移动电话或蜂巢式电话进行通信时, 也无可避免的会 使得电话非常靠近他的听力辅助器。
故在现今的通信科技中, 需要一种没有上述缺点但又能有效降低移动电 话或蜂巢式电话对音频设备所造成的脉沖干扰的方法及其装置, 申请人鉴于 已知技术中的缺失, 经过悉心试验与研究, 本着锲而不舍的精神构思出本发 明一一降低音频干扰的方法及其装置, 能够克服上述缺点, 以下为本发明的 筒要说明。 发明内容
为克服背景技术一节中所述已知技术所存在的缺陷, 本发明提出一种降 低噪声脉冲对音频设备的干扰的方法及其装置, 这些噪声脉冲是由移动电话 或蜂巢式电话等干扰源所产生的, 其概念主要是先找出由移动电话或蜂巢式 电话所造成的噪声脉沖的样板脉沖, 再将该样板脉沖从受到该噪声脉沖所影 响而含有该噪声脉沖波形的该音频设备的音频中予以减除, 然后将处理过的 音频以内插法予以 ~正, 如自回归内插 (auto-regressive interpolation), 然后再 视状况, 对修正后的音频施予一噪声抑制技术, 如频谱减除技术等噪声抑制 技术, 以上步骤在实施时, 将视该噪声脉冲的持续时间而分类为短突发 (short burst)或长突发 (long burst)而有不同可能的组合。
根据本发明的构思, 提出一种降低音频干扰的方法, 该音频受到一个或 一连串相似的噪声脉沖的千扰, 该噪声脉沖包括一中央脉沖及一延伸脉冲, 其特征在于:产生一样板脉沖以代表该噪声脉沖;及实施以下的步骤 (A)、(B)、 (C)或其组合以抑制该噪声脉沖:(A)从该音频中该噪声脉沖的位置减除该样板 脉冲或该样板脉冲中的该中央脉沖而产生一第一处理音频; (B)至少跨越该噪 声脉沖的该中央脉冲而内插该第一处理音频或该音频而产生的二第二处理音 频; 或 (C)以一噪声抑制技术处理该第二处理音频或该音频。
较佳地, 本发明所提供的该种降低音频干扰的方法, 其特征在于将该音 频分割为多个区间如帧, 并在每一个帧中实施抑制该噪声脉沖的步骤。
较佳地, 本发明所提供的该种降低音频千扰的方法, 其特征在于将一个 或连续多个噪声脉沖按其持续时间分类为长突发或短突发。
较佳地, 本发明所提供的该种降低音频干扰的方法, 其特征在于: 若所 述噪声脉沖为该短突发时, 则实施步骤 (A)、 (B)或其组合。
较佳地, 本发明所提供的该种降低音频干扰的方法, 其特征在于: 若所 述噪声脉冲为该长突发时, 则实施步骤 (A)、 (B)、 (C)或其组合。
较佳地, 本发明所提供的该种降低音频干扰的方法, 其中该产生步骤还 包括步骤: 确认该噪声脉冲在该音频中的一位置及该噪声脉冲的一振幅, 藉 此根据该振幅计算对该样板脉冲的缩放比例, 然后在该音频的相应位置上减 去该样板脉冲或该样板脉冲的该中央脉冲。
较佳地, 本发明所提供的该种降低音频干扰的方法, 其特征在于该产生 步骤还包括步骤: 估计该噪声脉冲所具有的多个特征由此产生该样板脉冲。
较佳地, 本发明所提供的该种降低音频干扰的方法, 其特征在于该内插 是利用一自回归模式, 一目前商业上可获得的恢复技术或其组合。
较佳地, 本发明所提供的该种降低音频干扰的方法, 其特征在于该噪声 抑制技术是一频谱减除技术,一目前商业上可获得的噪声抑制技术或其组合。
较佳地, 本发明所提供的该种降低音频千扰的方法, 其特征在于该噪声 脉冲是由一移动电话或一蜂巢式电话所产生。
较佳地, 本发明所提供的该种降低音频千扰的方法, 其特征在于该音频 是一音频设备所发出的音频, 该音频设备为一麦克风、 一音响、 一电话、 一 录音机、 一可携式音频播放器或一听力辅助器。
一种实施如前述的该种降低音频干扰的方法的装置。
根据本发明的构思, 提出一种降低音频干扰的装置, 其特征在于, 包括: 一输入, 用以接收一音频; 一存储器, 用以产生^ /或存储一噪声脉沖的一样 板脉冲, 其中该样板脉冲代表一连串相似的噪声脉沖的其中之一, 该噪声脉 沖包括一中央脉冲与一延伸脉冲; 及一噪声抑制器, 与该输入连接, 用以实 施以下步骤 (A;)、 (B)、 (C)或其组合以抑制该噪声脉沖: (A)从该音频中减除该 样板脉冲而产生一第一处理音频; (B)至少跨越该噪声脉沖的该中央脉冲而内 插该第一处理音频或该音频而产生一第二处理音频; ^/或 (C)以一噪声抑制技 术处理该第二处理音频或该音频。
较佳地, 本发明所提供的该种降低音频干扰的装置, 其特征在于还包括 一侦测器, 用以确认一噪声脉冲的出现。
较佳地, 本发明所提供的该种降低音频千扰的装置, 其特征在于所述噪 声抑制器还包括一振幅决定装置, 用以决定该脉沖的振幅。
较佳地, 本发明所提供的该种降低音频干扰的装置, 其特征在于所述噪 声抑制器还包括一脉冲位置侦测器用以侦测该噪声脉冲的一位置, 且该脉沖 位置侦测器还包括一交互相关侦测器。
小结, 本发明通过将该噪声脉冲分解为一中央脉冲与一延伸脉沖的方式 将该噪声脉沖的影响最小化, 其中该中央脉沖为来自于该蜂巢式电话或其他 干扰来源的电磁辐射产物, 该延伸脉沖则来自于该音频设备, 特别是该音频 设备的电容 (capacitance)特性的特征。 不过在将该中央脉冲样板化并减去该已 样板化中央脉沖时, 并未考虑该音频设备的音频电路特性。 附图概述
通过本发明实施例作为示例的方式, 并参考其伴随附图, 本发明的以上 与其他特征及优点由以下叙述将更加明显, 其中:
图 1 为一典型由移动电话或蜂巢式电话等干扰源所产生的噪声脉沖的示 意图;
图 1 为用以从音频中抑制噪声脉沖的一降低音频干扰的装置; 图 3 为本发明所提出的一种降低音频干扰的方法的实施流程图; 图 4 为分类为短突发的噪声脉冲示意图;
图 5 为分类为长突发的噪声脉冲示意图; 以及
图 6 为噪声脉沖的频谱示意图。 本发明的较佳实施方式
请参阅图 1, 图 1 为一典型的由移动电话或蜂巢式电话等干扰源所产生 的噪声脉冲的示意图, 图 1 中所示的噪声脉冲 τ包括两个主要部分, 一中央 脉沖 τι以及一延伸脉沖 τ2。 产生图 1 中所示的噪声脉沖的状况为, 当一音频 设备, 如: 麦克风、 音响、 电话、 录音机、 可携式音频播放器或听力辅助器 正在操作时, 也就是正在接收音频或播放音频时, 在这些音频设备周围正好 有移动电话或蜂巢式电话也正在进行通信或信息传输, 此时在音频设备中原 本是规则的或平滑的一正常音频, 将因为受到移动电话或蜂巢式电话所产生 的脉冲干扰, 而开始在原本的正常音频中夹杂着如图 1 中所示的噪声脉沖, 包括中央脉沖 τι以及延伸脉沖 τ2
如图 1 中所示的噪声脉沖 τ, 其中央脉沖 τι是由移动电话或蜂巢式电话 在进行通信或信息传输时所产生的电磁波所激发, 而延伸脉沖 τ2则与音频设 备的电容特性有关。 因此, 噪声脉沖 τ的特性将与该蜂巢式电话的电磁波放 射强度及音频设备的音频电路的特性有关, 移动电话或蜂巢式电话所传输的 音频脉冲的长度随着电话的规格而不同,一般来说对于 44千赫兹的采样频率 而言等于 25个样本,脉沖的振幅尤其与该蜂巢式电话与该音频设备之间的距 离成正比。
请继续参阅图 2, 为用来从音频中抑制噪声脉沖的一降低音频干扰的装 置。 如同在图 2中所显示的降低音频干扰的装置 20, 此降低音频干扰的装置 20的一输入 10与一音频电路 1连接,此音频电路 1为设置于各种音频设备中 的音频电路, 此音频设备或音频电路 1受到周围由移动电话或蜂巢式电话的 电磁脉冲所引起的干扰, 如果该音频为模拟形式, 该音频电路 1 的输出被提 供给一模拟数字转换器 2, 如果该音频已经为数字形式, 例如来自于一光盘, 则该模拟数字转换器 2将被省略, 且该音频电路 1的输出将被直接提供给一 先进先出 (FIFO)緩存器 3; 该模拟数字转换器 2 的输出被提供给该先进先出 (FIFO)緩存器 3, 而该先进先出 (FIFO)緩存器 3的输出则连接至一数字信号处 理器 (Digital Signal Processor)4; 一硬件电磁脉沖侦测器 5具有一连接至该数 字信号处理器 4的输出。 该模拟数字转换器 2也转换一输出至一存储器 6, 该存储器 6也可以由该数字信号处理器 4所存取。 该数字信号处理器 4的一 输出也被提供至一数字模拟转换器 7,该数字模拟转换器 7连接至一输出 11, 该输出 11将与某些如扬声器或耳机等换能器形式结合。 事实上, 也可将在图 2中所揭露的模拟数字转换器 2、 先进先出 (FIFO)緩存器 3、 数字信号处理器 4及数字模拟转换器 7等的组合, 视为一个滤波器 22或一个噪声抑制器 24, 其可移除存在于输入 10中的噪声脉沖, 而使得输出 11的音频为无噪声脉沖 干扰的音频。 其中数字信号处理器 4同时兼作为振幅决定装置用以决定该脉 沖的振幅, 及作为脉沖位置侦测器用以侦测该噪声脉沖的位置。
请继续参阅图 3, 为本发明所提出一种降低音频干扰方法的实施流程图。 在图 3中所揭露的该种降低音频千扰的方法, 其包括: 步骤 30, 将音频分割 为多个帧; 步骤 32, 产生一样板脉冲以代表该噪声脉冲; 步骤 34, 将该噪声 脉沖分类为一短突发或一长突发; 步骤 36, 若该噪声脉沖为该长突发时, 实 施步骤 (A)、 (B)、 (C)或其组合; 步骤 38, 若该噪声脉冲为该短突发时, 实施 步骤 (A)、 (B)或其组合; 步骤 (A)为从该音频中减除该样板脉冲而产生一第一 处理音频; 步骤 (B)至少跨越该噪声脉冲的该中央脉冲而内插该第一处理音频 或该音频而产生一第二处理音频; 步骤 (C)以一噪声抑制技术处理该第二处理 音频或该音频。 其中, 图 3中以虚线 (dashed line)表示的步骤, 如步骤 30、 34 及 38为可选性步骤,使用者在实施本发明所提出的一种降低音频干扰的方法 时, 可以自由选择步骤 30、 34及 38的实施与否。
实施以上方法时, 首先从正在操作中的音频设备获得一个音频, 此音频 设备因受到周围正在使用中或通信中的移动电话或蜂巢式电话的脉冲干扰, 使得音频中含有噪声脉冲, 此噪声脉沖分为两个部分, 分别为中央脉冲及延 伸脉冲, 其中中央脉冲是由移动电话或蜂巢式电话在进行通信或信息传输时 所产生的电磁波脉冲所激发, 而延伸脉冲则与音频设备的电容特性有关。
接着决定是否釆用基于帧 (frame-based)的方式来实施以上方法,若拟采用 基于帧的方式, 则须先将音频信号分割为若干个帧, 每个帧大约是 40/wsec长 度。 若不拟采用基于帧的方式来实施以上方法, 可忽略以上所述的方法。 以 上所述即为图 3 中所示的步骤 30。 以下以决定采用基于帧 (frame-based)的方 式来实施上述方法进行说明。
接着利用以下的方式产生一个足以代替此噪声脉沖的一个样板脉冲, 为 了构建代表由移动电话或蜂巢式电话所产生典型脉沖的一样板脉沖, 图 2中 所示的装置被设定为具有一零数值音频, 此时只有通过该音频电路 1并提供 至该数字信号处理器 4的信号为一蜂巢式电话脉沖, 其表示在正常操作中会 干扰正常音频的那些噪声脉沖, 计算出该脉沖的振幅、 周期等特征, 由此产 生一个样板脉沖并存储在存储器 6中。 当该装置在正常使用时, 根据振幅计 算对样板脉沖的缩放比例, 便可将样板脉沖经过比例化放大或缩小后而使其 与噪声脉冲相符, 噪声脉沖的中央脉沖与延伸脉沖的长度对于一已知规格的 蜂巢式电话与一已知规格的音频装置而言为常数, 但噪声脉沖的强度 (振幅) 将与移动电话或蜂巢式电话与该音频设备的距离有关。 该数字信号处理器 4 则被编程以利用在该存储器 6中所存储的样板脉沖进行在此所叙述的方法, 并同时产生如叙述的样板脉沖, 并在需要的时候将其更新。
当然还有其他的方式用以建立一样板脉沖, 如直接使用一个外部的侦测 器以估计样板脉冲、 使用历史经验所建立的样板脉冲、 使用前一次的样板脉 沖、 使用当音频设备中的音频为空白时所测得的噪声脉冲作为样板脉冲、 使 用前一帧所获得的噪声脉沖作为样板脉冲等方式。 但请特别注意, 建立样板 脉沖时, 并不一定需要在接收一音频的步骤之后, 若是采用如使用历史经验 所建立的样板脉冲、 使用前一次的样板脉冲、 使用当音频设备中的音频为空 白时所测得的噪声脉冲作为样板脉沖等方式获得样板脉沖, 则在接收一音频 步骤之前即可执行。 以上所述即为图 3中所示的步骤 32。
在帧中搜寻噪声脉沖并辨识每一噪声脉冲的起点。 若发现一个以上的噪 声脉沖, 决定噪声脉沖在帧中的出现频率并据此将此帧中的噪声脉沖分类为 长突发 (long burst)或短突发 (short burst) , 通常将持续超过 30 sec的噪声脉沖 分类为长突发, 未持续超过 30/«sec的噪声脉沖则分类为短突发。 请参阅图 4 和图 5, 分别为分类为短突发的噪声脉沖示意图及分类为长突发的噪声脉沖 示意图。
若此帧中的噪声脉沖分类为短突发, 则实施步骤 (A)、 (B)或步骤 (A)及 (B) 的组合步骤 (步骤 (A)及 (B)将于后文详述), 即, 依次实施步骤 (A)、 (B), 依次 实施步骤 (B)、 (A), 仅实施步骤 (A), 仅实施步骤 (B)等方式。 通常仅实施步骤 (A)或 (B)的其中一个步骤即可有效的抑制分类为短突发的噪声脉沖,特别是当 噪声脉沖的延伸脉沖相当明显时, 即可略去步骤 (B)的实施。
若此帧中的噪声脉沖分类为长突发, 则实施步骤 (A)、(B)、(C)或步骤 (A)、 (B)及 (C)的组合步骤 (步骤 (A)、 (B)及 (C)将于后文详述), 即, 可选择仅施实施 步骤 (A)、 (B)或 (C)的其中一个步骤或其中的两个步骤, 且在实施时, 步骤的 先后次序可依实际状况予以交替。 另外一种常见的状况为: 若此帧中的噪声 脉沖已分类为长突发, 则在基于短突发的步骤实施完毕后, 建议应再实施步 骤 (C), 或先行实施步骤 (C), 再实施基于短突发的步骤。
本发明使用以下三个步骤 (A)、 (B)、 (C)或 (A)、 (B)及 (C)的组合来抑制音 频中所携有的噪声脉沖。
首先说明步骤 (A),从该音频中每一噪声脉冲的位置减除该样板脉沖而产 生一第一处理音频的方式。 步骤 (A)中是利用以下所述的方程式移除音频中所 包含的由蜂巢式电话所造成的噪声脉冲:
xno,se(") = b{n)ypulse{n) + xoriginal{n) + en (1)
其中:
xnoise为含有由该蜂巢式电话所引起的噪声脉沖的音频信号;
为该蜂巢式电话的噪声脉沖的已知样板脉冲;
e„为一白噪声 (white noise)输人;
^■original为该原始不被干扰的音频。
如果已知该蜂巢式电话干扰在音频中的精确位置, 并可以决定该尺度因 子 b的话, 则可以尺度因子 b为准, 对样板脉沖 《 进行比例放大或縮小, 然后在音频的相应位置上减去样板脉冲或板脉冲的中央脉沖, 当样板脉冲 ypulse由含有噪声脉沖的音频信号 X„。 中减除后, 即可以较佳地复原该原始信 号 x0 al。 此原始信号; C。„.g,.„fl/也可暂时将其视为第一处理音频。 以上所述即 为图 3中所示的步骤 (A;)。
当进行步骤 (A)时, 会有需要知道噪声脉沖在音频中相应的精确位置, 然 后才能在此位置上减去已经过比例缩放后的样板脉冲或样板脉沖的中央脉 沖, 通常有许多侦测方法可用来决定该噪声脉沖的精确位置, 如: 1)硬件电 磁波侦测器, 2)信号振幅的最大门槛侦测, 3)信号斜率的最大门槛侦测, 4) 交互相关 /相符滤波侦测器, 5)贝氏阶梯侦测器, 以及 6)自回归侦测器等的装 置。 以上并未列举出所有的侦测器, 但对于实时应用而言, 硬件电磁波侦测 器可能是最方便及最有效的解决方式。 使用信号振幅最大门槛侦测^或信号 斜率最大门槛侦测的软件实现也是简单、 有效率及有效杲的解决方式。
接着说明步骤 (B), 至少跨越该噪声脉沖的该中央脉沖而内插该第一处理 音频或该音频而产生一第二处理音频。 对噪声脉冲而言, 通常由移动电话或 本), < ^因此可以利用 ^"个简单的内插模式假设在这、 25 ^个样本之间的信息是遗 失的, 并且进行内插。
假设在方程式 (1)中的尺度函数 b(«)可以被简化为对于所有的 "都具有一 固定值 b, 且已确认该噪声脉沖的位置, 则此时可利用如单一通道自回归模 式 (single channel auto-regressive (AR) model), 以一低阶次 (典型上 p=20)复原 / 内插该中央脉沖的位置, 以内插值取代中央脉沖的值:
^noise (") = Σ i oise (" - 0 + (2) 其中:
。 为包含该蜂巢式电话干扰的观测噪声信号:
e„为一白噪声输入。
对于单一通道最小平方自回归 (LSAR)的内插而言, 方程式 (2)的解答为:
为通过内插所求得的第二处理音频。
实际上, 步骤 (B)中所釆用的内插不仅仅限于 AR模式, 其他各种商业可 获得用于恢复音频的内插技术, 均可应用于步骤 (B)中, 模拟因中央脉冲的发 生所遗失的音频。
接着说明步骤 (C), 以一噪声抑制技术处理该第二处理音频或该音频。 步 骤 (C)中所指出的噪声抑制技术为目前及未来在商业上可获得的噪声抑制技 术, 如: 频瞽减除 (spectrum subtraction)技术、 功率傳减除 (power spectrum subtraction)技术以及其他目前及未来在商业上可获得的噪声抑制技术的组 合。 请参阅图 6, 为噪声脉沖的频谱示意图, 可通过以下方式建立每一帧中 的频谱或功率谱, 如直接使用一个外部的侦测器以估计频谱、 使用历史经验 所建立的频谱、 使用前一次的频谱、 使用当音频设备中的音频为空白时所测 得的频博作为样板频谘、 使用前一帧所获得的频谱等方式。
上述方程式 (1)和 (2)也可应用至多通道的音频, 用以抑制多通道音频所携 有的噪声脉沖。

Claims

权 利 要 求 书
1、 一种降低音频千扰的方法,该音频受到一个或一连串相似的噪声脉 冲的干扰, 该噪声脉沖包括一中央脉沖及一延伸脉冲, 其特征在于:
产生一样板脉沖以代表该噪声脉冲; 及
实施以下的步骤 (A)、 (B)、 (C)或其组合以抑制该噪声脉沖:
(A) 从该音频中该噪声脉冲的位置减除该样板脉冲或该样板脉冲中 的该中央脉沖而产生一第一处理音频;
(B) 至少跨越该噪声脉沖的该中央脉沖而内插该第一处理音频或该 音频而产生一第二处理音频; SJ或
(C) 以一噪声抑制技术处理该第二处理音频或该音频。
2、 根据权利要求 1所述的方法, 其特征在于:
将该音频分割为多个区间, 并在每一个区间中实施抑制该噪声脉冲的步 骤。
3、 根据权利要求 2所述的方法, 其特征在于该区间为一帧。
4、 根据权利要求 1所述的方法, 其特征在于:
将一个或连续多个噪声脉冲按其持续时间分类为长突发或短突发。
5、 根据权利要求 4所述的方法, 其特征在于:
若所述噪声脉沖为该短突发时, 则实施步骤 (A)、 (B)或其组合。
6、 根据权利要求 4所述的方法, 其特征在于:
若所述噪声脉沖为该长突发时, 则实施步骤 (A)、 (B)、 (C)或其组合。
7、 根据权利要求 1所述的方法, 其特征在于该产生步骤还包括步骤: 确认该噪声脉沖在该音频中的一位置及该噪声脉沖的一振幅, 藉此根据 该振幅计算对该样板脉冲的缩放比例, 然后在该音频的相应位置上减去该样 板脉沖或该样板脉沖的该中央脉沖。
8、 根据权利要求 1所述的方法, 其特征在于该产生步骤还包括步骤: 估计该噪声脉沖所具有的多个特征由此产生该样板脉沖。
9、 根据权利要求 1所述的方法, 其特征在于:
该内插是利用一自回归模式,一目前商业上可获得的恢复技术或其组合, 而该噪声抑制技术是一频谱减除技术, 一目前商业上可获得的噪声抑制技术 或其组合。
10、 根据权利要求 1所述的方法, 其特征在于:
该噪声脉沖是由一移动电话或一蜂巢式电话所产生, 而该音频是一音频 设备所发出的音频, 该音频设备为一麦克风、 一音响、 一电话、 一录音机、 一可携式音频播放器或一听力辅助器。
11、 一种实施根据权利要求 1所述的该种降低音频干扰的方法的装置。
12、 一种降低音频干扰的装置, 其特征在于, 包括:
一输入, 用以接收一音频;
一存储器, 用以产生及 /或存储一噪声脉冲的一样板脉沖, 其中该样板脉 沖代表一连串相似的噪声脉沖的其中之一, 该噪声脉冲包括一中央脉沖与一 延伸脉沖; 及
一噪声抑制器, 与该输入连接, 用以实施以下步骤 (A)、 (B)、 (C)或其组 合以抑制该噪声脉冲:
(A)从该音频中减除该样板脉冲而产生一第一处理音频; (B) 至少跨越该噪声脉冲的该中央脉沖而内插该第一处理音频或该 音频而产生一第二处理音频; H
(C) 以一噪声抑制技术处理该第二处理音频或该音频。
13、 根据权利要求 12所述的装置, 其特征在于还包括一侦测器, 用以 确认该噪声脉沖的出现。
14、 根据权利要求 12所述的装置, 其特征在于所述噪声抑制器还包括 一振幅决定装置, 用以决定该脉冲的振幅。
15、 根椐权利要求 12所述的装置, 其特征在于所述噪声抑制器还包括 一脉冲位置侦测器, 用以侦测该噪声脉沖的一位置, 且该脉沖位置侦测器还 包括一交互相关侦测器。
PCT/CN2008/000980 2008-05-21 2008-05-21 降低音频干扰的方法及其装置 WO2009140792A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2008/000980 WO2009140792A1 (zh) 2008-05-21 2008-05-21 降低音频干扰的方法及其装置
CN2008801293036A CN102037664A (zh) 2008-05-21 2008-05-21 降低音频干扰的方法及其装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2008/000980 WO2009140792A1 (zh) 2008-05-21 2008-05-21 降低音频干扰的方法及其装置

Publications (1)

Publication Number Publication Date
WO2009140792A1 true WO2009140792A1 (zh) 2009-11-26

Family

ID=41339715

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2008/000980 WO2009140792A1 (zh) 2008-05-21 2008-05-21 降低音频干扰的方法及其装置

Country Status (2)

Country Link
CN (1) CN102037664A (zh)
WO (1) WO2009140792A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019071127A1 (en) * 2017-10-05 2019-04-11 iZotope, Inc. IDENTIFICATION AND DELETION OF NOISE IN AN AUDIO SIGNAL

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH11154000A (ja) * 1997-09-16 1999-06-08 Denso Corp 雑音抑圧装置及び該装置を用いた音声認識システム
CA2326948A1 (en) * 2000-11-24 2002-05-24 Tet Hin Yeap Noise/interference suppression system
CN1192359C (zh) * 1998-12-18 2005-03-09 艾利森电话股份有限公司 用于抑制通信系统中噪声效应的方法和系统
CN1225104C (zh) * 2000-04-08 2005-10-26 阿尔卡塔尔公司 抑制噪声信号的方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH11154000A (ja) * 1997-09-16 1999-06-08 Denso Corp 雑音抑圧装置及び該装置を用いた音声認識システム
CN1192359C (zh) * 1998-12-18 2005-03-09 艾利森电话股份有限公司 用于抑制通信系统中噪声效应的方法和系统
CN1225104C (zh) * 2000-04-08 2005-10-26 阿尔卡塔尔公司 抑制噪声信号的方法
CA2326948A1 (en) * 2000-11-24 2002-05-24 Tet Hin Yeap Noise/interference suppression system

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019071127A1 (en) * 2017-10-05 2019-04-11 iZotope, Inc. IDENTIFICATION AND DELETION OF NOISE IN AN AUDIO SIGNAL
US11017795B2 (en) 2017-10-05 2021-05-25 iZotope, Inc. Identifying and addressing noise in an audio signal

Also Published As

Publication number Publication date
CN102037664A (zh) 2011-04-27

Similar Documents

Publication Publication Date Title
US8972251B2 (en) Generating a masking signal on an electronic device
CN107636758B (zh) 声学回声消除系统和方法
US8355511B2 (en) System and method for envelope-based acoustic echo cancellation
EP2966643B1 (en) Adaptive noise canceling system with source audio leakage detection
KR101444100B1 (ko) 혼합 사운드로부터 잡음을 제거하는 방법 및 장치
CN101826892B (zh) 回声消除器
US20160086594A1 (en) Noise reduction audio reproducing device and noise reduction audio reproducing methods
TWI543149B (zh) 雜訊消除方法
KR20150008472A (ko) 잡음 제거 개인용 오디오 디바이스에서 2차 경로 적응형 응답의 잡음 버스트 적응
JP2009503568A (ja) 雑音環境における音声信号の着実な分離
CN104751839A (zh) 具有低速率仿真的噪声消除系统
JP5834948B2 (ja) 残響抑制装置、残響抑制方法及び残響抑制用コンピュータプログラム
CN110956975B (zh) 回声消除方法及装置
US20150341722A1 (en) Methods and devices for reverberation suppression
US20120197635A1 (en) Method for generating an audio signal
CN112509595A (zh) 音频数据处理方法、系统及存储介质
JP2008309955A (ja) ノイズサプレス装置
CN106297816B (zh) 一种回声消除的非线性处理方法和装置及电子设备
JP2003284183A (ja) 反響抑圧装置、反響抑圧方法及びプログラム
CN108605191B (zh) 异音检测方法和装置
WO2023051622A1 (zh) 提升远场语音交互性能的方法和远场语音交互系统
WO2009140792A1 (zh) 降低音频干扰的方法及其装置
CN109994104B (zh) 一种自适应通话音量控制方法及装置
JPH07111527A (ja) 音声の加工方法およびそれを用いた装置
JP6480303B2 (ja) 無線装置

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 200880129303.6

Country of ref document: CN

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 08757326

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 08757326

Country of ref document: EP

Kind code of ref document: A1