WO2019047134A1 - 一种高生物拟真性语音处理滤波器与语音识别设备 - Google Patents

一种高生物拟真性语音处理滤波器与语音识别设备 Download PDF

Info

Publication number
WO2019047134A1
WO2019047134A1 PCT/CN2017/100975 CN2017100975W WO2019047134A1 WO 2019047134 A1 WO2019047134 A1 WO 2019047134A1 CN 2017100975 W CN2017100975 W CN 2017100975W WO 2019047134 A1 WO2019047134 A1 WO 2019047134A1
Authority
WO
WIPO (PCT)
Prior art keywords
transconductance amplifier
inverting input
filter
output
transconductance
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.)
Ceased
Application number
PCT/CN2017/100975
Other languages
English (en)
French (fr)
Inventor
张金勇
王磊
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.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
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 Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Priority to PCT/CN2017/100975 priority Critical patent/WO2019047134A1/zh
Publication of WO2019047134A1 publication Critical patent/WO2019047134A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208—Noise filtering
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
    • H03H17/00—Networks using digital techniques
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
    • H03H17/00—Networks using digital techniques
    • H03H17/02—Frequency selective networks

Definitions

  • the invention relates to the field of speech recognition, in particular to a high bio-realistic speech processing filter and a speech recognition device.
  • the key part of human perception of sound is the cochlea of the ear.
  • the cochlear perception of sound is different for different frequencies.
  • the frequency distribution from the cusp to the volute is from about 20 Hz to 20 kHz.
  • the biological cochlear has a specific frequency response curve to the sound.
  • the current speech processing filter basically adopts an ordinary analog band pass filter or a digital filter, and does not consider the sound frequency response characteristic of the biological cochlea.
  • the object of the present invention is to provide a high bio-realistic speech processing filter and a speech recognition device, which aim to solve the problem that when the existing analog band pass filter or digital filter performs speech processing on human voice, the recognition effect is poor and easy. Distortion problem.
  • the present invention provides a high bio-realistic speech processing filter, the filter being a nine-order filter constructed by a cascade of a band pass filtering unit, a low pass filtering unit, and an elliptical low pass filtering unit.
  • the center frequency of the ninth order filter is adjusted by the low pass filtering unit.
  • the high bio-realistic speech processing filter provided by the invention adopts advanced low-power analog integrated circuit technology, combined with the frequency response characteristic of the biological cochlear, the sound response curve near a specific center frequency can be subdivided into a relatively slow passive segment, The active selective phase and the abrupt steep transition phase are treated with bandpass filtering in the slower passive segments and low-pass filtering in the active selective phase, with sudden abrupt changes.
  • an elliptical low-pass filter unit is used for processing, thereby realizing speech processing with high bio-realism and low power consumption.
  • FIG. 1 is a schematic block diagram of a high biological immersive speech processing filter according to an embodiment of the present invention
  • Figure 2 is a frequency response curve of the cochlea to sound
  • FIG. 3 is a circuit schematic diagram of a transconductance amplifier according to an embodiment of the present invention.
  • FIG. 4 is a schematic circuit diagram of a band pass filtering unit of the high-reality immersive speech processing filter shown in FIG. 1;
  • FIG. 5 is a schematic diagram of a low-pass filter unit circuit of the high-reality immersive speech processing filter shown in FIG. 1;
  • FIG. 6 is a schematic circuit diagram of an elliptical low-pass filter unit of the high-reality immersive speech processing filter shown in FIG. 1;
  • FIG. 7 is a high frequency 12 KHz frequency response simulation curve of the high bio-realistic speech processing filter shown in FIG. 1;
  • FIG. 8 is a low frequency 20 Hz frequency response simulation curve of the high bio-realistic speech processing filter shown in FIG. 1.
  • FIG. 8 is a low frequency 20 Hz frequency response simulation curve of the high bio-realistic speech processing filter shown in FIG. 1.
  • the high bio-realistic speech processing filter in the embodiment of the present invention can be applied to a speech recognition device such as a cochlear implant, a hearing aid, etc.
  • the high bio-realistic speech processing filter is a band pass filtering unit 10 (Band Pass Filter, BPF), low pass filter unit 20 (Low Pass Filter, LPF) and elliptical low pass filter unit 30 (Elliptic Filter, ELF) are sequentially cascaded to construct a 9th order filter, wherein the 9th order filter The center frequency can be adjusted by the low pass filtering unit 20.
  • the sound response curve for a particular center frequency can be subdivided into three phases, the first being a slower passive phase, where the bandpass filtering unit 10 is used for processing, and the second is the active selective phase. At this time, the processing is performed using the low-pass filter unit 20, and the third is the cut-off phase of sudden abrupt change, at which time the processing is performed using the elliptical low-pass filter unit 30.
  • the nine-order filter adopts a G m -C form, wherein G m is a transconductance value of a transconductance (Operation) amplifier (OTA), and C is a capacitance.
  • G m is a transconductance value of a transconductance (Operation) amplifier (OTA), and C is a capacitance.
  • OTA transconductance
  • the G m -C filter path constituent unit mainly has a transconductance amplifier and a capacitor.
  • the transconductance operational amplifier includes a first POMS tube M1, a second PMOS transistor M2, a third PMOS transistor M3, a fourth PMOS transistor M4, a first NMOS transistor M5, and a second NMOS.
  • the gate of the first POMS tube M1 and the gate of the second PMOS transistor M2 are connected to the voltage source VDD.
  • the source of the first POMS tube M1 and the source of the second PMOS tube M2 are connected to the adjustable bias current ISS.
  • the center frequency of the corresponding filter can be changed by changing the magnitude of the bias current ISS.
  • the drain of the first POMS tube M1 is connected to the source of the third PMOS transistor M3, the drain of the second PMOS transistor M2 is connected to the source of the fourth PMOS transistor M4, and the gate of the third PMOS transistor M3 is used as a transconductance operational amplifier.
  • the inverting input terminal has a drain of the third PMOS transistor M3 connected to the drain of the first NMOS transistor M5 and serves as a non-inverting input terminal of the transconductance operational amplifier.
  • the source of the first NMOS transistor M5, the drain of the third NMOS transistor M7, the gate of the third NMOS transistor M7, and the gate of the fourth NMOS transistor M8 are connected in common, and the source of the third NMOS transistor M7 is grounded;
  • the drain of the PMOS transistor M2 is connected to the source of the fourth PMOS transistor M4, and the fourth PMOS
  • the gate of transistor M4 acts as the output of the transconductance operational amplifier.
  • the drain of the fourth PMOS transistor M4 is connected in common with the drain of the second NMOS transistor M6, the gate of the second NMOS transistor M6, and the gate of the first NMOS transistor M5.
  • the source of the second NMOS transistor M6 is connected to the drain of the fourth NMOS transistor M8, and the drain of the fourth NMOS transistor M8 is grounded.
  • the transfer function of the band pass filtering unit 10 is expressed as:
  • the transfer function of the low pass filtering unit 20 is expressed as:
  • the transfer function of the elliptical low pass filtering unit 30 is expressed as:
  • ⁇ 0 is the center frequency of the ninth-order filter
  • s represents the complex field
  • ⁇ is the gain variable
  • the transconductance value G m ⁇ 0 C 0 of each transconductance amplifier in the band pass filtering unit 10, where ⁇ 0 is the center frequency of the ninth order filter, and C 0 is the capacity of the reference capacitor.
  • the band pass filtering unit 10 includes a first transconductance amplifier Gm1, a second transconductance amplifier Gm2, a third transconductance amplifier Gm3, a first capacitor C1, and a second capacitor C2.
  • the non-inverting input terminal of the first transconductance amplifier Gm1 is connected to the initial input signal V i
  • the inverting input terminal of the first transconductance amplifier Gm1 is connected to the inverting input terminal of the second transconductance amplifier Gm2
  • the first transconductance amplifier Gm1 The output terminal is connected to the non-inverting input terminal of the second transconductance amplifier Gm2 and grounded through the first capacitor C1
  • the inverting input terminal of the second transconductance amplifier Gm2 is connected to the output terminal thereof, and is grounded through the second capacitor C2, the third transconductance
  • the inverting input terminal of the amplifier Gm3 is connected to the reference voltage signal V ref
  • the non-inverting input terminal of the third transconductance amplifier Gm3 is connected to the initial input signal V i
  • the output end of the third transconductance amplifier Gm3 is connected to the second transconductance amplifier Gm2
  • the capacity of the first capacitor C1 and the second capacitor C2 is C 0
  • the low pass filtering unit 20 includes two low pass filters 201 connected in series, and the low pass filter 201 includes a fourth transconductance amplifier Gm4, a fifth transconductance amplifier Gm5, a third capacitor C3, and a fourth capacitor. C4.
  • the non-inverting input terminal of the fourth transconductance amplifier Gm4 is connected to the input signal, that is, the non-inverting input terminal of the fourth transconductance amplifier Gm4 of the low-pass filter 201 of the preceding stage is connected to the primary filtered signal V 0 , and the fourth transconductance amplifier Gm4
  • the inverting input terminal is connected to the inverting input terminal of the fifth transconductance amplifier Gm5, and the output end of the fourth transconductance amplifier Gm4 is connected to the non-inverting input terminal of the fifth transconductance amplifier Gm5 and grounded through the third capacitor C3, the fifth span
  • the output terminal of the lead amplifier Gm5 is connected to its inverting input terminal, and is grounded through the fourth capacitor C4.
  • the output end of the fifth transconductance amplifier Gm5 serves as the output end of the low pass filter 201, that is, the secondary low pass filter 201.
  • the output of the fifth transconductance amplifier Gm5 outputs a secondary filtered
  • the capacity of the third capacitor C3 and the fourth capacitor C4 in the low pass filter 201 is 0.6667C 0
  • the elliptical low pass filtering unit 30 includes a first elliptical filter 301 and a second elliptical filter 302 connected in series.
  • the first elliptical filter 301 includes a sixth transconductance amplifier Gm6, a seventh transconductance amplifier Gm7, a fifth capacitance C5, a sixth capacitance C6, and a seventh capacitance C7.
  • the non-inverting input of the sixth transconductance amplifier Gm6 is connected to the output signal of the low-pass filter unit 20 (secondary filtered signal V 1 ), and the inverting input terminal of the sixth transconductance amplifier Gm6 is connected to the inversion of the seventh transconductance amplifier Gm7.
  • the output of the fifth transconductance amplifier Gm5 is connected to the non-inverting input terminal of the seventh transconductance amplifier Gm7 and grounded through the fifth capacitor C5, and the output end of the seventh transconductance amplifier Gm7 is connected to its inverting input terminal, and
  • the sixth transconductance amplifier Gm6 is connected to the output terminal of the seventh transconductance amplifier Gm7 through the seventh capacitor C7, and the output terminal of the seventh transconductance amplifier Gm7 is connected to the second elliptic filter 302. Input.
  • the capacities of the fifth capacitor C5 and the sixth capacitor C6 are respectively 0.3016C 0 and 1.467C 0 .
  • the second elliptical filter 302 includes an eighth transconductance amplifier Gm8, a ninth transconductance amplifier Gn9, and an eighth capacitance C8.
  • the forward input terminal of the eighth transconductance amplifier Gm8 is connected to the output signal of the first elliptical filter 301, and the inverting input terminal of the eighth transconductance amplifier Gm8 is connected to the reference voltage signal V ref , and the output terminal of the eighth transconductance amplifier Gm8 is connected.
  • the inverting input terminal and the output terminal of the ninth transconductance amplifier Gn9, the forward input terminal of the ninth transconductance amplifier Gn9 is connected to the reference voltage signal V ref , and the ninth transconductance amplifier Gn9 is grounded through the eighth capacitor C8 and serves as an ellipse
  • the output of the low pass filtering unit 30 outputs the cubic filtered signal V 2 .
  • the capacity of the eighth capacitor C8 is 1.27C 0 .
  • the signal passes through the BPF, LPF, and ELF in sequence.
  • ⁇ B , ⁇ L and ⁇ E represent the natural frequencies of BPF, LPF and ELF, respectively
  • Q B and Q L represent the quality factors of BPF and LPF, respectively
  • ELF is 1 dB according to the pass band ripple. Designed with a third-order low-pass elliptical filter with 40dB attenuation.
  • the above filter transfer function related parameters are set based on the latest physiological experiment data.
  • Q B is set to 1.
  • ⁇ L and ⁇ E are set to 1.5 ⁇ B according to the center frequency position of the biological cochlear frequency response curve.
  • ⁇ B is represented by ⁇ 0
  • its size can be changed between 20Hz and 20KHz
  • Q L is represented by 1/ ⁇ , and its value can be performed according to sound stimulation of different intensities. Adjust to determine the filter quality factor for various sound intensities.
  • the ninth-order filter frequency of the embodiment of the present invention is adjustable from 20 Hz to 12 kHz
  • the gain at the center frequency ⁇ 0 is adjustable
  • the gain range at 20 Hz is from 1.431 dB to 31.81 dB
  • the quality factor Q is The range is from 4.635 to 11.54
  • the gain range is from 6.495dB to 34.05dB at 12KHz
  • the quality factor Q ranges from 0.48 to 14.96
  • the cutoff frequency is above 300dB/dec, even up to 392.39dB/dec, which is in good agreement with the biocochlear response curve. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Signal Processing (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Quality & Reliability (AREA)
  • Computer Hardware Design (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Networks Using Active Elements (AREA)

Abstract

一种高生物拟真性语音处理滤波器，滤波器为由带通滤波单元（10）、低通滤波单元（20）及椭圆低通滤波单元（30）依次级联构建的九阶滤波器，其中，九阶滤波器的中心频率通过低通滤波单元（20）来调整。这样，结合生物耳蜗的频率响应特性，特定中心频率附近的声音响应曲线可以细分为较缓的无源段、有源的具有选择性的阶段以及突然陡变的截止阶段，较缓的无源段采用带通滤波单元（10）进行处理，有源的具有选择性的阶段采用低通滤波单元（20）进行处理，突然陡变的截止阶段采用椭圆低通滤波单元（30）进行处理，从而实现具有高生物拟真性、功耗低的语音处理。

Description

一种高生物拟真性语音处理滤波器与语音识别设备 技术领域
本发明涉及语音识别领域，特别是涉及一种高生物拟真性语音处理滤波器与语音识别设备。
背景技术
人对声音感知的关键部位是耳朵的耳蜗，耳蜗对声音的感知不同部位针对不同的频率，从蜗尖到蜗底响应频率分布范围由约20Hz到20KHz。此外，根据已有科学研究表明，生物耳蜗对声音有特定的频率响应曲线，目前的语音处理滤波器基本采用普通的模拟带通滤波器或者数字滤波器，没有考虑生物耳蜗的声音频率响应特性。
人的语音本身是连续时域的，数字滤波的形式首先需要把模拟信号转换成数字信号，根据奈奎斯特采样定律，转换过程很容易出现数字冗余和失真现象，而且数字滤波很难做到低功耗，滤波速度也比较缓慢。采用普通的模拟带通滤波器完全没有考虑生物耳蜗的频率响应特性，会损失很多关键的语音信息，从而会降低人声音的识别效果。
发明内容
本发明的目的在于提供一种高生物拟真性语音处理滤波器与语音识别设备，旨在解决对现有模拟带通滤波器或者数字滤波器对人声音进行语音处理时，存在识别效果差、容易失真的问题。
一方面，本发明提供了一种高生物拟真性语音处理滤波器，所述滤波器为由带通滤波单元、低通滤波单元及椭圆低通滤波单元依次级联构建的九阶滤波器，所述九阶滤波器的中心频率通过所述低通滤波单元来调整。
另一方面，本发明还提供了一种语音识别设备，包括上述高生物拟真性语音处理滤波器。
本发明提供的高生物拟真性语音处理滤波器采用先进的低功耗模拟集成电路技术，结合生物耳蜗的频率响应特性，特定中心频率附近的声音响应曲线可以细分为较缓的无源段、有源的具有选择性的阶段以及突然陡变的截止阶段，在较缓的无源段采用带通滤波进行处理，在有源的具有选择性的阶段则采用低通滤波进行处理，在突然陡变的截止阶段则采用椭圆低通滤波单元进行处理，从而实现具有高生物拟真性、低功耗的语音处理。
附图说明
图1为本发明实施例中高生物拟真性语音处理滤波器的模块示意图；
图2为耳蜗对声音的频率响应曲线；
图3为本发明实施例中跨导放大器的电路原理图；
图4为图1所示开高生物拟真性语音处理滤波器的带通滤波单元电路原理图；
图5为图1所示开高生物拟真性语音处理滤波器的低通滤波单元电路原理图；
图6为图1所示开高生物拟真性语音处理滤波器的椭圆低通滤波单元电路原理图；
图7为图1所示高生物拟真性语音处理滤波器的高频12KHz频率响应仿真曲线；以及
图8为图1所示高生物拟真性语音处理滤波器的低频20Hz频率响应仿真曲线。
具体实施方式
为了使本发明要解决的技术问题、技术方案及有益效果更加清楚明白，以 下结合附图及实施例，对本发明进行进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。
请参阅图1和图2，本发明实施例中的高生物拟真性语音处理滤波器可应用在人工耳蜗、助听器等语音识别设备，该高生物拟真性语音处理滤波器为由带通滤波单元10(Band Pass Filter，BPF)、低通滤波单元20(Low Pass Filter，LPF)及椭圆低通滤波单元30(Elliptic Filter，ELF)依次级联构建的九阶滤波器，其中，该九阶滤波器的中心频率可通过低通滤波单元20来调整。如此，针对特定中心频率附近的声音响应曲线可以细分为三个阶段，第一是较缓的无源阶段，此时使用带通滤波单元10进行处理，第二是有源的具有选择性阶段，此时使用低通滤波单元20进行处理，第三是突然陡变的截止阶段，此时使用椭圆低通滤波单元30进行处理。
在本发明实施例中，九阶滤波器采用Gm-C形式，其中Gm为跨导(运算)放大器(OTA)的跨导值，C为电容。Gm-C滤波器路组成单元主要有跨导放大器和电容。
请参阅图3，在本发明实施例中，跨导运算放大器包括第一POMS管M1、第二PMOS管M2、第三PMOS管M3、第四PMOS管M4、第一NMOS管M5、第二NMOS管M6、第三NMOS管M7、第四NMOS管M8。
第一POMS管M1的栅极和第二PMOS管M2的栅极共接电压源VDD，第一POMS管M1的源极和第二PMOS管M2的源极共接可以调整的偏置电流ISS，通过改变偏置电流ISS的大小可以改变相应滤波器的中心频率。第一POMS管M1的漏极接第三PMOS管M3的源极，第二PMOS管M2的漏极接第四PMOS管M4的源极，第三PMOS管M3的栅极作为跨导运算放大器的反相输入端，第三PMOS管M3的漏极接第一NMOS管M5的漏极并作为跨导运算放大器的正相输入端。第一NMOS管M5的源极、第三NMOS管M7的漏极、第三NMOS管M7的栅极及第四NMOS管M8的栅极共接，第三NMOS管M7的源极接地；第二PMOS管M2的漏极接第四PMOS管M4的源极，第四PMOS 管M4的栅极作为跨导运算放大器的输出端。第四PMOS管M4的漏极与第二NMOS管M6的漏极、第二NMOS管M6的栅极和第一NMOS管M5的栅极共接。第二NMOS管M6的源极接第四NMOS管M8的漏极，第四NMOS管M8的漏极接地。
在本发明实施例中，带通滤波单元10的传递函数表示为：
Figure PCTCN2017100975-appb-000001
低通滤波单元20的传递函数表示为：
Figure PCTCN2017100975-appb-000002
椭圆低通滤波单元30的传递函数表示为：
Figure PCTCN2017100975-appb-000003
由此可得，九阶滤波器的传递函数为：
Figure PCTCN2017100975-appb-000004
其中，ω0为九阶滤波器的中心频率，s表示复数域，β为增益变量。
带通滤波单元10中各个跨导放大器的跨导值Gm＝ω0C0，其中，ω0为九阶滤波器的中心频率，C0为基准电容的容量。请参阅图4，带通滤波单元10包括第一跨导放大器Gm1、第二跨导放大器Gm2、第三跨导放大器Gm3、第一电容C1和第二电容C2。
第一跨导放大器Gm1的正相输入端接入初始输入信号Vi，第一跨导放大器Gm1的反相输入端接第二跨导放大器Gm2的反相输入端，第一跨导放大器Gm1的输出端接第二跨导放大器Gm2的正相输入端且通过第一电容C1接地，第二跨导放大器Gm2的反相输入端接其输出端，并通过第二电容C2接地，第三跨导放大器Gm3的反相输入端接入基准电压信号Vref，第三跨导放大器Gm3的正相输入端接初始输入信号Vi，第三跨导放大器Gm3的输出端接第二跨导放大器 Gm2的输出端，且第二跨导放大器Gm2的输出端作为带通滤波单元10的输出端输出一次滤波信号V0。在本发明实施例中，第一电容C1和第二电容C2的容量为C0，第一跨导放大器Gm1、第二跨导放大器Gm2和第三跨导放大器Gm3的跨导值Gm1＝Gm2＝Gm3＝ω0C0。
请参阅图5，低通滤波单元20包括两个串联连接的低通滤波器201，低通滤波器201包括第四跨导放大器Gm4、第五跨导放大器Gm5、第三电容C3和第四电容C4。
第四跨导放大器Gm4的正相输入端接输入信号，即前级的低通滤波器201的第四跨导放大器Gm4的正相输入端接入一次滤波信号V0，第四跨导放大器Gm4的反相输入端接第五跨导放大器Gm5的反相输入端，第四跨导放大器Gm4的输出端接第五跨导放大器Gm5的正相输入端且通过第三电容C3接地，第五跨导放大器Gm5的输出端并接其反相输入端，且通过第四电容C4接地，第五跨导放大器Gm5的输出端作为低通滤波器201的输出端，即次级的低通滤波器201的第五跨导放大器Gm5的输出端输出二次滤波信号V1。
在本发明实施例中，第四跨导放大器Gm4的跨导值
Figure PCTCN2017100975-appb-000005
第五跨导放大器Gm5的跨导值Gm5＝βω0C0，其中，C0为基准电容的容量。β出现在LPF的传输函数中表示增益变量，通过调整β这个参数可以调整滤波器的增益，具体可以调节跨导放大器的偏置电流ISS的大小，如此，以调整九阶滤波器的中心频率。在本发明实施例中，低通滤波器201中的第三电容C3和第四电容C4的容量为0.6667C0
椭圆低通滤波单元30中各个跨导放大器的跨导值Gm＝ω0C0。请参阅图6，椭圆低通滤波单元30包括串联连接的第一椭圆滤波器301和第二椭圆滤波器302。
第一椭圆滤波器301包括第六跨导放大器Gm6、第七跨导放大器Gm7、第五电容C5、第六电容C6和第七电容C7。第六跨导放大器Gm6的正相输入端 接低通滤波单元20的输出信号(二次滤波信号V1)，第六跨导放大器Gm6的反向输入端接第七跨导放大器Gm7的反相输入端，第五跨导放大器Gm5的输出端接第七跨导放大器Gm7的正相输入端且通过第五电容C5接地，第七跨导放大器Gm7的输出端并接其反相输入端，且通过第六电容C6接地，第六跨导放大器Gm6的正相输入端通过第七电容C7接第七跨导放大器Gm7的输出端，第七跨导放大器Gm7的输出端接第二椭圆滤波器302的输入端。在本发明实施例中，第五电容C5、第六电容C6的容量分别为0.3016C0、1.467C0。
第二椭圆滤波器302包括第八跨导放大器Gm8、第九跨导放大器Gn9和第八电容C8。第八跨导放大器Gm8的正向输入端接第一椭圆滤波器301的输出信号，第八跨导放大器Gm8的反相输入端接基准电压信号Vref，第八跨导放大器Gm8的输出端接第九跨导放大器Gn9的反相输入端和输出端，第九跨导放大器Gn9的正向输入端接基准电压信号Vref，第九跨导放大器Gn9的通过第八电容C8接地，且作为椭圆低通滤波单元30的输出端输出三次滤波信号V2。在本实施例中，第八电容C8的容量为1.27C0。
在滤波器Gm-C电路中，信号依次通过BPF、LPF、ELF。在滤波器传输函数中ωB、ωL和ωE分别代表BPF、LPF和ELF的固有频率，QB和QL分别代表BPF和LPF的品质因子，ELF则根据通带纹波为1dB、阻带衰减为40dB的三阶低通椭圆滤波器进行设计。
上述滤波器传输函数相关参数根据最新生理实验数据来设置。在强声音刺激情况下，从生物耳蜗的频率响应曲线可知，QB设为1比较适合，在微弱声音刺激时，根据生物耳蜗频率响应曲线中心频率位置，ωL和ωE设置为1.5ωB，ωB用ω0来表示，其大小可以在20Hz到20KHz之间改变，确定整个滤波器通道的中心频率位置，QL用1/β来表示，其值可以根据不同强度的声音刺激来进行调节，以确定各种声音强度情况下的滤波器品质因子大小。
请参阅图7和图8，其中，本发明实施例的九阶滤波器频率从20Hz到12KHz 可调，中心频率ω0处增益可调，20Hz处增益范围从1.431dB到31.81dB，品质因子Q范围从4.635到11.54，12KHz处增益范围从6.495dB到34.05dB，品质因子Q范围从0.48到14.96，此外截止频率都在300dB/dec以上，甚至高达392.39dB/dec，与生物耳蜗响应曲线非常吻合。
以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等，均应包含在本发明的保护范围之内。

Claims (14)

  1. 一种高生物拟真性语音处理滤波器，其特征在于，所述滤波器为由带通滤波单元、低通滤波单元及椭圆低通滤波单元依次级联构建的九阶滤波器，所述九阶滤波器的中心频率通过所述低通滤波单元来调整。
  2. 如权利要求1所述的高生物拟真性语音处理滤波器，其特征在于，所述九阶滤波器采用Gm-C形式，其中Gm为跨导放大器的跨导值，C为电容。
  3. 如权利要求1所述的高生物拟真性语音处理滤波器，其特征在于，所述带通滤波单元的传递函数表示为：
    Figure PCTCN2017100975-appb-100001
    其中，ω0为所述中心频率，s表示复数域。
  4. 如权利要求1-3任一所述的高生物拟真性语音处理滤波器，其特征在于，所述带通滤波单元中各个跨导放大器的跨导值Gm＝ω0C0，其中，ω0为所述中心频率，C0为基准电容的容量。
  5. 如权利要求4所述的高生物拟真性语音处理滤波器，其特征在于，所述带通滤波单元包括第一跨导放大器、第二跨导放大器、第三跨导放大器、第一电容和第二电容，其中：
    所述第一跨导放大器的正相输入端接入初始输入信号，所述第一跨导放大器的反相输入端接所述第二跨导放大器的反相输入端，所述第一跨导放大器的输出端接所述第二跨导放大器的正相输入端且通过所述第一电容接地，所述第二跨导放大器的反相输入端接其输出端，并通过所述第二电容接地，所述第三跨导放大器的反相输入端接入所述基准电压信号，所述第三跨导放大器的正相输入端接初始输入信号，所述第三跨导放大器的输出端接所述第二跨导放大器的输出端，且所述第二跨导放大器的输出端作为所述带通滤波单元的输出端。
  6. 如权利要求1所述的高生物拟真性语音处理滤波器，其特征在于，所述低通滤波单元的传递函数表示为：
    Figure PCTCN2017100975-appb-100002
    其中，ω0为所述中心频率，s表示复数域，β为增益变量。
  7. 如权利要求1、2或6所述的高生物拟真性语音处理滤波器，其特征在于，所述低通滤波单元包括两个串联连接的低通滤波器。
  8. 如权利要求7所述的高生物拟真性语音处理滤波器，其特征在于，所述低通滤波器包括第四跨导放大器、第五跨导放大器、第三电容和第四电容，其中：
    所述第四跨导放大器的正相输入端接输入信号，所述第四跨导放大器的反相输入端接所述第五跨导放大器的反相输入端，所述第四跨导放大器的输出端接所述第五跨导放大器的正相输入端且通过所述第三电容接地，所述第五跨导放大器的输出端并接其反相输入端，且通过所述第四电容接地，所述第五跨导放大器的输出端作为所述低通滤波器的输出端。
  9. 如权利要求8所述的高生物拟真性语音处理滤波器，其特征在于，所述第四跨导放大器的跨导值
    Figure PCTCN2017100975-appb-100003
    所述第五跨导放大器的跨导值Gm5＝βω0C0，其中，ω0为所述中心频率，C0为基准电容的容量，β为增益变量。
  10. 如权利要求1所述的高生物拟真性语音处理滤波器，其特征在于，所述椭圆低通滤波单元的传递函数表示为：
    Figure PCTCN2017100975-appb-100004
    其中，ω0为所述中心频率，s表示复数域。
  11. 如权利要求10所述的高生物拟真性语音处理滤波器，其特征在于，所述椭圆低通滤波单元包括串联连接的第一椭圆滤波器和第二椭圆滤波器。
  12. 如权利要求11所述的高生物拟真性语音处理滤波器，其特征在于，所述第一椭圆滤波器包括第六跨导放大器、第七跨导放大器、第五电容、第六电容和第七电容，其中：
    所述第六跨导放大器的正相输入端接所述低通滤波单元的输出信号，所述第六跨导放大器的反向输入端接所述第七跨导放大器的反相输入端，所述第五跨导放大器的输出端接所述第七跨导放大器的正相输入端且通过所述第五电容接地，所述第七跨导放大器的输出端并接其反相输入端，且通过所述第六电容接地，所述第六跨导放大器的正相输入端通过所述第七电容接所述第七跨导放大器的输出端，所述第七跨导放大器的输出端接所述第二椭圆滤波器。
  13. 如权利要求12所述的高生物拟真性语音处理滤波器，其特征在于，所述第二椭圆滤波器包括第八跨导放大器、第九跨导放大器和第八电容，其中：
    所述第八跨导放大器的正向输入端接所述第一椭圆滤波器的输出信号，所述第八跨导放大器的反相输入端接基准电压信号，所述第八跨导放大器的输出端接所述第九跨导放大器的反相输入端和输出端，所述第九跨导放大器的正向输入端接所述基准电压信号，所述第九跨导放大器的通过所述第八电容接地，且作为椭圆低通滤波单元的输出端。
  14. 一种语音识别设备，其特征在于，包括权利要求1至13任一项所述高生物拟真性语音处理滤波器。
PCT/CN2017/100975 2017-09-08 2017-09-08 一种高生物拟真性语音处理滤波器与语音识别设备 Ceased WO2019047134A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2017/100975 WO2019047134A1 (zh) 2017-09-08 2017-09-08 一种高生物拟真性语音处理滤波器与语音识别设备

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2017/100975 WO2019047134A1 (zh) 2017-09-08 2017-09-08 一种高生物拟真性语音处理滤波器与语音识别设备

Publications (1)

Publication Number Publication Date
WO2019047134A1 true WO2019047134A1 (zh) 2019-03-14

Family

ID=65634705

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/100975 Ceased WO2019047134A1 (zh) 2017-09-08 2017-09-08 一种高生物拟真性语音处理滤波器与语音识别设备

Country Status (1)

Country Link
WO (1) WO2019047134A1 (zh)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN2105140U (zh) * 1991-11-05 1992-05-20 国营成都宏明无线电器材总厂 有源滤波器
JP2002156997A (ja) * 2000-11-21 2002-05-31 Sharp Corp 音声検出制御装置
JP2005128132A (ja) * 2003-10-22 2005-05-19 Matsushita Electric Ind Co Ltd 話速変換方法及び話速変換装置
CN101656079A (zh) * 2009-07-20 2010-02-24 霍桂馨 一种便携式汉语声调学习机的语音处理方法
CN107799122A (zh) * 2017-09-08 2018-03-13 中国科学院深圳先进技术研究院 一种高生物拟真性语音处理滤波器与语音识别设备

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN2105140U (zh) * 1991-11-05 1992-05-20 国营成都宏明无线电器材总厂 有源滤波器
JP2002156997A (ja) * 2000-11-21 2002-05-31 Sharp Corp 音声検出制御装置
JP2005128132A (ja) * 2003-10-22 2005-05-19 Matsushita Electric Ind Co Ltd 話速変換方法及び話速変換装置
CN101656079A (zh) * 2009-07-20 2010-02-24 霍桂馨 一种便携式汉语声调学习机的语音处理方法
CN107799122A (zh) * 2017-09-08 2018-03-13 中国科学院深圳先进技术研究院 一种高生物拟真性语音处理滤波器与语音识别设备

Similar Documents

Publication Publication Date Title
Lyon et al. History and future of auditory filter models
Kumngern et al. 0.5 V sixth-order Chebyshev band-pass filter based on multiple-input bulk-driven OTA
CN107799122B (zh) 一种高生物拟真性语音处理滤波器与语音识别设备
CN102497168A (zh) 智能音量限制装置及方法
Kumngern et al. Voltage-mode multifunction biquadratic filters using new ultra-low-power differential difference current conveyors
CN105048989A (zh) 具有共模噪声消减的低通滤波器
Kumngern et al. 31.3 nW, 0.5 V bulk-driven OTA for biosignal processing
CN108769851B (zh) 一种数字耳机
JP6750901B2 (ja) チャンネル選択フィルタを備えた高線形性ＷｉＧｉｇベースバンドアンプ
CN203596803U (zh) 一种频率补偿的装置
WO2019047134A1 (zh) 一种高生物拟真性语音处理滤波器与语音识别设备
CN108462479B (zh) 基于改进型Gm-C的镜像抑制滤波器及其构建方法
CN106658299B (zh) 音频处理电路及终端设备
CN106249793B (zh) 电源稳压器电路
CN113965197A (zh) 一种rtr分频方法及电路
CN102395077B (zh) 一种抗干扰耳机
Wang et al. A floating active inductor based CMOS cochlea filter with high tunability and sharp cut-off
Wei et al. A computationally efficient non-uniform digital FIR filter bank for hearing aid
CN206894870U (zh) 音频解码装置及其音频输出电路
CN204859532U (zh) 一种模拟类麦克风回音降噪压缩处理电路
CN203912166U (zh) 一种有源带阻滤波电路
CN108235189A (zh) 一种语音信号的回声消除装置及电视机
CN115664369A (zh) 一种基于自偏置伪电阻的可调低通滤波器电路
Maghami et al. Compact, programmable, two-Stage configuration for implantable biopotential recording amplifiers
Kambalimath et al. FPGA based implementation of comb filters for use in binaural hearing aids for reducing intraspeech spectral masking

Legal Events

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

Ref document number: 17924582

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: 17924582

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 22.09.2020)

122 Ep: pct application non-entry in european phase

Ref document number: 17924582

Country of ref document: EP

Kind code of ref document: A1