JP2008301427A - Multichannel voice reproduction equipment - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide multichannel voice reproduction equipment capable of improving voice clarity in content such as a movie by emphasizing speech voice included mainly in a front center signal, and further making a listener feel as that the speech voice is reproduced in a listening position. <P>SOLUTION: The multichannel voice reproduction equipment includes a speech extracting means for outputting a speech extraction signal of the front center signal, a speech emphasizing means for providing speech emphasizing process to the speech extraction signal and outputting as a speech emphasizing signal, and a crosstalk canceling means for branching the speech emphasizing signal into a first speech emphasizing signal and a second speech emphasizing signal, providing crosstalk canceling process for eliminating interaural crosstalk of the listener on the respective signals, and generating a first crosstalk cancel signal and second crosstalk cancel signal. The first crosstalk cancel signal is reproduced from a left speaker, and the second crosstalk cancel signal is reproduced from a right speaker. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a multi-channel sound reproducing apparatus that reproduces multi-channel sound including surround sound, and in particular, emphasizes speech sound mainly included in a front center signal to improve sound clarity in content such as a movie. The present invention relates to an improved multi-channel audio reproduction apparatus.

  There is provided a multi-channel audio reproduction apparatus that reproduces a surround sound field by reproducing a so-called multi-channel audio signal having a plurality of independent audio signal channels exceeding two together with a video signal or by an audio signal alone. . Various systems and formats have been proposed, including the speaker arrangement direction, playback frequency band, surround system, etc. corresponding to each channel. In many cases, the multi-channel audio signal may include a signal of a total of five channels including a front center signal C, a surround left signal SL, and a surround right signal SR together with the front left signal L and the front right signal R. is there. In addition to these, there is a case where it is configured by a so-called 5.1 channel including a bass signal LFE.

  The multi-channel sound is scheduled to be reproduced from a speaker group including a plurality of speakers connected to the multi-channel sound reproducing device and arranged corresponding to each channel. A front audio signal (including a front left signal L, a front right signal R, and a front center signal C. The same applies hereinafter) is a front speaker (a front left speaker Lsp, a front right speaker Rsp, a front center) arranged in front of the viewer. It is reproduced from the speaker Csp. Surround sound signals (including a surround left signal SL and a surround right signal SR; the same applies hereinafter) include respective surround speakers (including a surround left speaker SLsp and a surround right speaker SRsp) arranged on the side or rear of the viewer. (The same shall apply hereinafter.)

  However, if the speaker group that plays back multi-channel audio consists of only the front speakers placed in front of the viewer and does not include surround speakers that play back only the surround audio signals, that is, if the speakers are minimal. When only the front left speaker Lsp and the front right speaker Rsp for general stereo reproduction are provided, the multi-channel audio reproducing apparatus is a stereo in which the front center signal and the surround audio signal are mixed into the front left signal and the front right signal. An audio signal is generated and reproduced from a front speaker. In the process of generating the stereo audio signal, in addition to the case of performing a downmix process (including a simple phase shift process) for adding the surround audio signal to the front audio signal, the head is applied to the surround audio signal. There is a case where a so-called virtual surround process is performed in which an audio signal that has been subjected to virtual localization processing based on a transfer function HRTF (Head Related Transfer Function) is added to the forward audio signal. If a stereo sound signal obtained by virtually localizing a surround sound signal into a front sound signal is not generated, sound waves of the surround sound signal arrive at the front left speaker Lsp and the front right speaker Rsp from the side or rear of the listener. This causes a problem that such a sound field cannot be reproduced.

  However, conventionally, there is a problem that in the case of content such as a movie having a wide dynamic range of a multi-channel audio signal, it is sometimes difficult to hear the speech of the characters. In content such as a movie, the speech is generally distributed to the front center signal of the multi-channel audio signal. Therefore, conventionally, in order to improve the intelligibility of speech speech, there is a technique in which enhancement processing is performed by increasing the level of a frequency region corresponding to a human speech band.

  For example, a frequency component extraction circuit for extracting a predetermined frequency component corresponding to a human voice from an audio signal accompanying an image is added, and a signal obtained by passing the left and right signals of a stereo signal through the extraction circuits is added. An audio circuit for a television receiver that inputs to the circuit and the first subtraction circuit, outputs the output of the addition circuit from the center speaker through the gain control circuit, and controls the gain control circuit by the output of the first subtraction circuit; Yes (Patent Document 1).

  Conventionally, the frequency characteristic of the audio signal of the channel including the audio signal of the specific frequency band among the multi-channel audio signals including at least the right channel and the left channel is corrected based on the head-related transfer function. Frequency characteristic correcting means for correcting according to the characteristics, and the corrected audio signal is mixed with the right channel audio signal and the left channel audio signal and output as a right channel output audio signal and a left channel output audio signal, respectively. And a multi-channel audio signal processing circuit (Patent Document 2).

JP-A-4-249484 (FIG. 1) JP 2004-266604 A (FIG. 1)

  However, in a multi-channel audio reproduction device, when performing a downmix process or a virtual surround process in which a surround audio signal is added to a front audio signal, the dynamic range tends to decrease because the signals are added. There is a limit to improving the speech intelligibility by the conventional method of changing the voice level as in the case. Even in the case of Patent Document 2 in which correction is performed according to the correction characteristic determined based on the head-related transfer function, this correction characteristic corrects the sound coming from the left and right directions to the viewer from the front direction. This is only a characteristic for correcting the direction of arrival of the voice felt by the viewer.

  The present invention has been made in order to solve the above-described problems of the prior art, and an object of the present invention relates to a multi-channel audio reproducing apparatus that reproduces multi-channel audio including surround audio, and particularly to a front center signal. A multi-channel audio reproduction device that improves speech intelligibility in content such as a movie by emphasizing the speech contained in the sound and further making it feel as if the speech is being reproduced at the listening position of the listener There is.

  The multi-channel sound reproduction apparatus of the present invention reproduces the speech sound signal included in the front center signal of the multi-channel sound from the left speaker and the right speaker installed in front of the listener, and the speech sound at the listener's listening position is reproduced. A multi-channel audio reproducing apparatus for emphasizing speech extraction means for outputting a speech extraction signal that has passed through the speech sound band of the front center signal, and a maximum level near the upper limit of the speech sound band of the speech extraction signal. Line emphasis processing for relatively increasing the level and outputting as a line emphasis signal, the line emphasis signal is branched into a first line emphasis signal and a second line emphasis signal, and each of them is A first crosstalk cancel signal and a second crosstalk cancel are performed by performing a crosstalk cancellation process for removing interaural crosstalk. Comprising a crosstalk canceling means for generating a Le signal, and a first crosstalk cancel signal reproduced from the left speaker to reproduce the second crosstalk canceling signal from the right speaker.

  More preferably, in the multi-channel sound reproducing apparatus, the line extracting means includes a line detecting means for detecting whether or not a line sound is included in the front center signal, and the line detecting means is one of envelope signals of the line extracted signal. When the output level of the speech extraction signal is controlled according to the level of the ~ 4 Hz component level, and when detecting that speech speech is included, the output level of the speech extraction signal is increased, and when detecting that speech speech is not included Decreases the output level of the dialogue extraction signal.

  Preferably, the multi-channel sound reproduction apparatus generates a speech removal front center signal obtained by subtracting the speech extraction signal from the front center signal, and the speech removal front center signal as the first and second crosstalk cancellation signals. And a speech removal front center signal and the first crosstalk cancellation signal are reproduced from the left speaker, and a speech removal front center signal and the second crosstalk cancellation signal are reproduced from the right speaker.

  Preferably, the multi-channel sound reproducing device includes speech removal means for generating a speech-removed front center signal obtained by subtracting the speech-extracted signal from the front center signal, and reproduces the speech-removed front center signal from the center speaker. A crosstalk cancellation signal is reproduced from the left speaker, and a second crosstalk cancellation signal is reproduced from the right speaker.

  Preferably, in the multi-channel audio reproducing device, when the crosstalk canceling unit inputs only one of the first speech enhancement signal and the second speech enhancement signal subjected to the crosstalk cancellation processing, The selection input means for selecting and inputting either of the case of inputting at the reverse homology level or the case of inputting both of them at the reverse homology level, to the listener by selection in either case of the selection input means If you feel speech in the vicinity of one of the listener's ears, if you feel speech in the listener's head, or if you feel speech in the back of the listener Switch to one of the following.

  Preferably, the multi-channel audio reproduction device is the multi-channel audio reproduction device further receiving the front left signal and the front right signal among the multi-channel audio, and adds the front left signal to the first crosstalk cancellation signal, And an adding means for adding the front right signal to the second crosstalk cancellation signal.

  Preferably, the multi-channel audio reproduction device is the multi-channel audio reproduction device that further receives a surround left signal and a surround right signal among the multi-channel audio, and transmits a virtual arrival angle to the surround left signal and the surround right signal. Crosstalk giving means for generating a first crosstalk giving signal and a second crosstalk giving signal by performing a crosstalk giving process for giving an interaural crosstalk based on a function; and a first line of the first crosstalk giving signal. Adding means for adding to the enhancement signal and adding the second crosstalk giving signal to the second speech enhancement signal.

  The operation of the present invention will be described below.

  The multi-channel sound reproduction apparatus of the present invention reproduces the speech sound signal included in the front center signal of the multi-channel sound from the left speaker and the right speaker installed in front of the listener, and the speech sound at the listener's listening position is reproduced. The multi-channel sound reproducing apparatus to be emphasized corresponds to a case of further receiving a front left signal and a front right signal and a case of further receiving a surround left signal and a surround right signal among the multi-channel sound. In addition, in addition to the left speaker and the right speaker, when the center speaker and each surround speaker arranged on the side or rear of the viewer are further provided, the multi-channel sound reproducing device includes the left speaker and the right speaker. The surround sound field can be reproduced from these speakers as well as from.

  The multi-channel audio reproduction device includes a line extraction unit that outputs a line extraction signal that has passed through the line audio band of the front center signal, and a relative level that is near the upper limit of the line audio band of the line extraction signal. A line emphasizing process that increases the line emphasis and outputs as a line emphasis signal, and the line emphasis signal is branched into a first line emphasis signal and a second line emphasis signal. Crosstalk cancellation means for generating a first crosstalk cancellation signal and a second crosstalk cancellation signal by performing a crosstalk cancellation process for removing the talk. Therefore, it is possible to improve the clarity of the speech that is distributed to the front center signal of the multi-channel audio signal and reproduce it. In addition, it is preferable that the line extraction means for outputting the line extraction signal allows a line voice band of 200 Hz to 5 kHz to pass.

  The line extraction means may include a line detection means for detecting whether or not the line speech is included in the front center signal according to the magnitude of the 1 to 4 Hz component level of the envelope signal of the line extraction signal. In addition, the multi-channel sound reproducing apparatus may further include a speech removing unit that generates a speech removed front central signal obtained by subtracting the speech extraction signal from the forward central signal. As a result, the multi-channel sound reproducing apparatus of the present invention can emphasize speech that is mainly included in the front center signal by the speech extraction means and speech enhancement means, and the sound other than the speech sound included in the forward center signal. Can be reproduced in the surround sound field without being lost by the dialogue-removed front center signal.

  Here, the crosstalk cancellation processing is a head related transfer function HRTF (Head) generated from a spatial arrangement of a left speaker and a right speaker installed in front of the listener and both ears (left ear and right ear) of the listener. Signal transfer based on Related Transfer Function), canceling the crosstalk of spatial propagation from the left speaker to the right ear, or from the right speaker to the left ear, for example, the playback sound from the left channel is only to the left ear This is signal processing for transmitting the reproduced sound from the right channel only to the right ear. The crosstalk cancellation process almost eliminates audio playback to the ears on the opposite side (the ears far from the speakers) in the listener's ears, so if the headphones are worn or if the speakers are near the pinna When arranged, the sound field close to is reproduced. Therefore, the listener can obtain a feeling that the reproduced sound source exists in the immediate vicinity of the ear by the crosstalk cancellation process.

  In the multi-channel audio reproduction device of the present invention, the first crosstalk cancellation signal based on the emphasized speech is reproduced from the left speaker, and the second crosstalk cancellation signal is reproduced from the right speaker. The listener perceives that speech is being played in the immediate vicinity of the ear. Therefore, in addition to the speech sound being emphasized by the speech emphasis process, the localization sound is different from the surround sound field other than the speech sound, and as a result, the intelligibility of the speech sound can be improved for the listener. it can.

  Since the first crosstalk cancellation signal and the second crosstalk cancellation signal are generated by performing the crosstalk cancellation process on the first dialogue enhancement signal and the second dialogue enhancement signal that branch the dialogue enhancement signal, the first dialogue enhancement signal And the second line emphasis signal, select either one of the two, the case where both are input at the same homology level, or the case where both are input at the reverse homology level. Input, it is possible to select and provide emphasized reproduction of speech to the listener. Specifically, speech speech enhancement to the listener is based on whether the speech sound is felt near one of the listener's ears, the speech sound is felt inside the listener's head, It is possible to switch to either the case where a speech voice is felt at the back of the head. Therefore, in addition to the emphasis process being performed only when the speech is present by the speech extraction means, the listener can select a sense of localization of the speech that is easy to hear, and the clarity of the speech is further increased. Can feel like.

  In addition, since the multi-channel audio reproduction device of the present invention includes the crosstalk canceling unit, the multichannel audio reproducing device further includes a crosstalk applying unit that applies interaural crosstalk based on the transfer function of the virtual arrival angle to the surround left signal and the surround right signal. By providing, virtual surround processing can be performed on the surround left signal and the surround right signal. Therefore, a surround sound field can be provided to the listener by a virtual sound source that is virtually localized laterally or rearward even without a surround speaker.

  The multi-channel audio reproduction apparatus of the present invention emphasizes the dialogue sound mainly included in the front center signal, and obtains a feeling that the reproduction sound source exists in the immediate vicinity of the listener's ear by the crosstalk cancellation process. Can do. By making the listener feel as if the speech is being played at the listening position of the listener, it is possible to improve the audio clarity in content such as a movie.

  The multi-channel audio reproduction device of the present invention is intended to improve the audio intelligibility in content such as a movie, and the multi-channel audio reproduction device outputs a dialogue extraction signal that passes the dialogue audio band of the front center signal. Extraction means, speech emphasis processing for relatively increasing the level so that the upper limit of the speech voice band of the speech extraction signal is at the maximum level, and outputting the speech emphasis signal as a speech emphasis signal; The first dialogue emphasis signal and the second dialogue emphasis signal are branched to each other, and the first crosstalk cancellation signal and the second crosstalk cancellation signal are obtained by performing a crosstalk cancellation process for removing the interaural crosstalk of the listener. Crosstalk canceling means for generating the first crosstalk cancellation signal from the left speaker. And, by configuring so as to reproduce the second crosstalk canceling signal from the right speaker, it was realized.

  Hereinafter, although the multi-channel audio | voice reproduction apparatus by preferable embodiment of this invention is demonstrated, this invention is not limited to these embodiment.

  FIG. 1 is a diagram for explaining a multi-channel audio reproducing apparatus 1 according to a preferred embodiment of the present invention. FIG. 1A is a block diagram illustrating a DSP (Digital Signal Processor) 10 that performs signal processing in the multi-channel audio reproduction apparatus 1, and FIG. 1B is a graph illustrating frequency characteristics of the dialogue enhancement processing. It is. The DSP 10 can change the signal processing by changing its processing program and coefficients. In FIG. 1A, configurations other than the DSP 10 of the multi-channel audio reproduction device 1 (for example, a power supply circuit, a control circuit, etc.) are omitted. In addition, a left speaker Lsp and a right speaker Rsp, which will be described later, are connected to the multichannel audio reproduction device 1 of the present embodiment.

  Specifically, the multi-channel audio reproduction device 1 includes a decoding circuit (not shown) that decodes multi-channel audio data input from another reproduction device (not shown) such as a disc player into a 5-channel surround audio signal. A D / A converter (not shown) converts the output signal from the DSP 10 into an analog signal, and an amplifier (not shown) that amplifies the output of the DA converter (not shown) and outputs it to a plurality of speakers (not shown) including. Note that the decoding circuit may be included in the DSP 10. The five-channel audio signal is a total of five channels including components of all bands of the front left signal L, the front right signal R, the front center signal C, the surround left signal SL, and the surround right signal SR, and is band-limited. The bass signal LFE or other surround channel signal may be further included.

  The DSP 10 of the multi-channel audio reproduction device 1 performs enhancement processing and crosstalk cancellation processing on the speech that is distributed to the front center signal C. The front center signal C is input to the line extraction circuit 11 and output as a line extraction signal Cd0. The speech extraction circuit 11 includes a band pass filter (BPF) 11a whose pass band is set to 200 Hz to 5 kHz corresponding to a human voice band. The speech extraction signal Cd0 is input to the speech enhancement circuit 12 and output as a speech enhancement signal Ce0. The line emphasis circuit 12 increases a frequency of 200 Hz to 5 kHz corresponding to a human voice band to a level relatively higher than other bands, and a band of about 4 kHz which is a band corresponding to a higher-order formant of human voice. Is increased to a relatively higher level, and is constituted by a band-pass filter (not shown). Therefore, the dialogue emphasis signal Ce0 is a signal with improved clarity of dialogue speech as shown in the frequency characteristics of FIG.

  The dialogue emphasis signal Ce0 is input to the selection input circuit 13. The selection input circuit 13 includes multipliers 13a and 13b that multiply the line emphasis signal Ce0 by coefficients ke1 and ke2, respectively. The outputs of the multipliers 13a and 13b are divided into two first line emphasis signals. The signal Ce1 and the second dialogue emphasis signal Ce2 are obtained. As will be described later, the coefficients ke1 and ke2 are obtained by inputting only one of the first dialogue emphasizing signal Ce1 and the second dialogue emphasizing signal Ce2, and inputting both at the same homology level. When switching between the case of inputting at the reverse homology level and the case of selecting and inputting one of the cases, one of values of 1, 0, and −1 is selected. The first speech enhancement signal Ce1 and the second speech enhancement signal Ce2 are input to the crosstalk cancellation circuit 14 via an adder 18 described later.

  The crosstalk cancellation circuit 14 receives the first speech enhancement signal Ce1 and the second speech enhancement signal Ce2 and outputs the first crosstalk cancellation signal Cc1 and the second crosstalk cancellation signal Cc2. The first crosstalk cancellation signal Cc1 is reproduced from the left speaker Lsp and the second crosstalk cancellation signal Cc2 is reproduced from the right speaker Rsp through an adder 15 and an amplifier (not shown) which will be described later. The first crosstalk cancellation signal Cc1 is included in the output signal Lv from the DSP 10, and the second crosstalk cancellation signal Cc2 is included in the output signal Rv from the DSP 10, and the left speaker Lsp and the right speaker Rsp, respectively. Is input.

  FIG. 2 is a diagram for explaining the arrangement of the front speakers in the surround sound field reproduced by the multi-channel sound reproducing device 1 and the head-related transfer function HRTF of the listener 2. The crosstalk cancellation circuit 14 of the multi-channel audio reproduction device 1 is composed of filters having transfer functions H11, H12, H21, and H22 calculated based on the head-related transfer functions h11, h12, h21, and h22. Lattice type filter. A lattice filter composed of filters having transfer functions H11, H12, H21, and H22 cancels the crosstalk of spatial propagation, and the head related transfer function h12 from the left speaker Lsp to the right ear 2R of the listener 2 or This is a filter that tries to make the head-related transfer function h21 from the right speaker Rsp to the left ear 2L of the listener 2 substantially close to 0 (zero). Note that the lattice type filter of the crosstalk cancellation circuit 14 may be replaced with a shuffler type filter including an adder, a subtracter, and two filters as long as the arrangement of the front speakers with respect to the listener 2 is symmetrical.

  Therefore, when the first crosstalk cancellation signal Cc1 and the second crosstalk cancellation signal Cc2 are reproduced from the left speaker Lsp and the right speaker Rsp, respectively, the speech included in the first speech enhancement signal Ce1 is the left ear 2L of the listener 2. The speech contained in the second speech enhancement signal Ce2 is reproduced to the right ear 2R of the listener 2. Since the multi-channel audio reproduction device 1 includes the crosstalk cancellation circuit 14, as a result, the listener 2 can use virtual speakers (CEL and CER indicated by dotted lines in FIG. 2) near the pinna of both ears. Can be obtained, and a sense that the emphasized speech is reproduced from these can be obtained.

  FIG. 3 is a diagram for explaining the sense of localization of the speech sound obtained by the listener 2 by the multi-channel sound reproducing device 1. The listener 2 obtains a feeling that the virtual speaker CEL is arranged in the vicinity of the left ear 2L of the listener 2 and the virtual speaker CER is arranged in the vicinity of the right ear 2R of the listener 2, so that the selection input circuit By changing the coefficients ke1 and ke2 of the 13 multipliers 13a and 13b, an image localization feeling as shown by the point collection region x in each of FIGS. 3 (a) to 3 (d) is realized. FIG. 3A shows a case where the coefficient of the selection input circuit 13 is (ke1 = 1, ke2 = 0) and the emphasized speech is localized in the vicinity of the left ear 2L. FIG. Is a case where the coefficient of the selection input circuit 13 is (ke1 = 0, ke2 = 1) and the emphasized speech is localized in the vicinity of the right ear 2R. FIG. 3C shows the selection input circuit. The coefficient of 13 is (ke1 = kex, ke2 = kex, where 0.5 ≦ kex ≦ 1), and the emphasized speech is localized in the head of the listener 2, FIG. d) is a case where the coefficients of the selection input circuit 13 are (ke1 = kex, ke2 = −kex, where 0.5 ≦ kex ≦ 1), and the emphasized speech is localized near the back of the listener 2 It is. The listener 2 can change the coefficients ke1 and ke2 of the selection input circuit 13 by operating the multi-channel sound reproducing device 1, and can select a localization that makes it easy to hear the emphasized speech.

  On the other hand, the DSP 10 of the multi-channel sound reproducing device 1 includes a dialogue removal circuit 16 that generates a dialogue removal front center signal Cf0 obtained by subtracting the dialogue extraction signal Cd0 from the front center signal C. The speech removal circuit 16 multiplies the speech extraction signal Cd0 by a coefficient (-1) and inverts the phase thereof, and subtracts the speech extraction signal Cd0 from the forward central signal C to obtain a speech removal forward central signal Cf0. An adder 16b to be generated and a multiplier 16c that multiplies the output of the adder 16b by a coefficient kc1 (= 0.707). The speech removal front center signal Cf 0 is input to the adder 15. The adder 15 adds the front left signal L, the first crosstalk cancellation signal Cc1, and the speech removal front center signal Cf0 to generate the output signal Lv, the front right signal R, and the second crosstalk cancellation signal. And an adder 15b that adds the Cc2 and the speech removal front center signal Cf0 to produce the output signal Rv.

  The speech removal front center signal Cf0 is reproduced from the front left speaker Lsp and the front right speaker Rsp at the same homologous level. Therefore, the listener 2 feels that the speech included in the front center signal C is reproduced from the virtual speakers CEL and CER near the pinna of both ears as described above, while the front center other than the speech It is possible to obtain a sense of localization such that the sound of the signal C (music components other than human speech, etc.) is reproduced from the virtual central speaker VC in front of the listener 2. The front left signal L is reproduced from the front left speaker Lsp via the adder 15, and the front right signal R is reproduced from the front right speaker Rsp via the adder 15. As a result, the listener 2 has the sound other than the speech localized in the speech reproduced from the front speaker, and only the emphasized speech speech is reproduced from the vicinity of the pinna of the listener's both ears. Therefore, the intelligibility of speech with different localization feeling is improved.

  Further, the surround left signal SL and the surround right signal SR are input to the crosstalk applying circuit 17 of the DSP 10. The crosstalk applying circuit 17 applies the head related transfer functions d11, d12, d21, and d22 assumed between the two ears 2L and 2R of the listener 2 and the virtual left surround speakers VSL and VSR to be virtually localized. It has a lattice type filter composed of filters having transfer functions D11, D12, D21 and D22 calculated based on them. The head-related transfer functions d12 and d21 mean crosstalk to the ear far from the virtual surround speaker. The outputs vs1 and vs2 of the lattice filter of the crosstalk applying circuit 17 are added to the first speech emphasis signal Ce1 and the second speech emphasis signal Ce2 via the adder 18 and input to the crosstalk cancel circuit 14. Note that the lattice type filter of the crosstalk giving circuit 17 may be replaced with a shuffler type filter including an adder, a subtracter, and two filters as long as the virtual surround speakers are symmetrically arranged with respect to the listener 2.

  Therefore, the surround left signal SL and the surround right signal SR are reproduced from the front left speaker Lsp and the front right speaker Rsp via the crosstalk giving circuit 17 and the crosstalk cancellation circuit 14, and the listener 2 It can be felt that surround sound is reproduced from the virtual left surround speakers VSL and VSR. As a result, even if the playback system including the multi-channel audio playback device 1 is not provided with a surround speaker, the listener 2 can receive a surround sound field by using a virtual sound source that is virtually localized laterally (or rearward). Can provide. Further, the multi-channel sound reproducing device 1 allows the listener 2 to reproduce only the emphasized speech sound from the vicinity of the ears of both ears of the listener 2 in the surround sound field reproduced from the front speaker. On the other hand, since the sound other than the speech is localized around the listener 2, the listener 2 can improve the intelligibility regarding the speech with different localization feelings.

  FIG. 4 is a diagram for explaining a multi-channel sound reproducing apparatus 3 according to another preferred embodiment of the present invention. The multi-channel sound reproducing apparatus 3 has the configuration of the DSP 10 that performs signal processing in the multi-channel sound reproducing apparatus according to the previous embodiment. Other than the playback apparatus 1 is common. In FIG. 4, configurations (for example, a power supply circuit, a control circuit, etc.) other than the DSP 10 of the multi-channel audio reproduction device 3 are omitted, and portions common to the first embodiment are denoted by the same reference numerals and described. Is omitted.

  FIG. 5 is a diagram for explaining the arrangement of the front speakers and the surround speakers in the surround sound field reproduced by the multi-channel sound reproducing device 3, and the head-related transfer function HRTF of the listener 2. In addition to the left speaker Lsp and the right speaker Rsp, the front center speaker Csp, the surround left speaker SLsp, and the surround right speaker SRsp are connected to the multi-channel sound reproducing device 3 of the present embodiment. The

  The DSP 10 of the multi-channel audio reproduction device 3 performs enhancement processing and crosstalk cancellation processing on the speech that is distributed to the front center signal C, as in the previous embodiment. The front center signal C is input to the line extraction circuit 11 and output as a line extraction signal Cd0. Since the emphasis processing for the speech and the crosstalk cancellation processing are the same as those in the previous embodiment, description thereof will be omitted. However, the speech extraction circuit 11 of this embodiment includes speech in the front center signal C in addition to the bandpass filter (BPF) 11a whose passband is set to 200 Hz to 5 kHz corresponding to the human speech band. The line detection circuit 11b detects whether or not the level of the 1 to 4 Hz component level of the envelope signal output from the band pass filter (BPF) 11a, and the level of the output from the band pass filter (BPF) 11a based on the envelope signal. And a multiplier 11c for controlling.

  FIG. 6 is a diagram for explaining the dialogue detection circuit 11b of this embodiment, FIG. 6 (a) is a block diagram showing the configuration of the dialogue detection circuit 11b, and FIG. 6 (b) is a dialogue detection circuit 11b. Is a flowchart illustrating an algorithm for generating a coefficient kz of the multiplier 11c output from The dialogue extraction circuit 11 operates adaptively depending on whether or not speech speech is included in the front center signal C, and when the speech detection circuit 11b detects that speech speech is included, the coefficient kz of the multiplier 11c. Is increased to increase the output level of the speech extraction signal Cd0, and when detecting that speech speech is not included, the coefficient kz of the multiplier 11c is decreased to decrease the output level of the speech extraction signal Cd0. it can.

  Since the output signal Xn of the band pass filter (BPF) 11a of the line extraction circuit 11 mainly includes a band component of 200 Hz to 5 kHz corresponding to the human voice band, the feature point of the line voice signal is detected. To determine if the signal is a line. Dialogue speech, which is human speech, generally has more intermittent elements than music signals, and its envelope is generally around a few Hz unless it is particularly fast. By extracting a few Hz component of the envelope, and determining whether the signal is speech or not depending on its level, the speech is extracted more appropriately.

  Specifically, as shown in FIG. 6A, the line detection circuit 11b first obtains the absolute value of the output signal Xn of the band pass filter (BPF) 11a, and extracts the envelope of the line sound signal (S101). ). Next, the 1 to 4 Hz component level Qn of the envelope signal is extracted by a low-pass filter (LPF: S102) having a cutoff frequency of 4 Hz and a high-pass filter (HPF: S103) having a cutoff frequency of 1 Hz. The level Yn of the envelope signal of several Hz is obtained by smoothing the absolute value of Qn (S104), and the coefficient kz of the multiplier 11c is controlled by the level Yn (S105).

  An algorithm shown in FIG. 6B for generating the coefficient kz of the multiplier 11c from the level Yn of the envelope signal of the speech sound signal will be described. First, the preset minimum level value Ymin is substituted into the register value Yp, and the coefficient kz is initialized to 0 (S201). When the level Yn of the envelope signal is input (S202), if the level Yn is greater than or equal to the register value Yp (S203: Yes), the level Yn is substituted for the register value Yp (S204) and the next step (S208) Proceed to On the other hand, when the level Yn is smaller than the register value Yp (S203: No), a value obtained by multiplying the register value Yp by the coefficient 0.9 is set as a new register value Yp (S205), and this register value Yp is the minimum level. When the value is less than or equal to the value Ymin (S206: Yes), the minimum level value Ymin is substituted into the register value Yp (S207), and the process proceeds to the next step (S208). When the register value Yp is larger than the minimum level value Ymin (S206: No), the process proceeds to the next step (S208).

  Next, when the register value Yp is greater than or equal to the preset maximum level value Ymax (S208: Yes), the maximum level value Ymax is substituted into the register value Yp, and the coefficient kz is set to 1 (S209). . On the other hand, when the register value Yp is smaller than the preset maximum level value Ymax (S208: No), if the minimum level value Ymin and the register value Yp are equal (S210: Yes), the coefficient kz is set to 0. Set (S211). If the minimum level value Ymin and the register value Yp are not equal (S210: No), (Yp−Ymin) / (Ymax−Ymin) is calculated and set to the coefficient kz (S212). Finally, the integer n is incremented as the level Yn of the next envelope signal is input (S213).

  As described above, the level Yn of the speech signal envelope signal included in the front center signal C is obtained, and the coefficient kz of the multiplier 11c can be controlled by the level Yn. In this way, the coefficient kz changes gently between 0 and 1 according to the level Yn. The algorithm for controlling the coefficient kz of the multiplier 11c is not limited to the above, and any algorithm that changes the coefficient kz relative to the fluctuation of the level Yn of the envelope signal may be used. Since the speech of a person can be produced substantially by detecting fluctuations in human speech, the clarity of the listener 2 can be significantly increased due to the difference in localization from other speech.

  Note that this adaptive processing for detecting speech speech has a low filter frequency and a long time constant, and it takes some time to detect whether speech speech is speech speech. Accordingly, in the dialogue detection circuit 11b of the dialogue extraction circuit 11 for the front center signal C, a delay circuit TD is provided in each signal path of multi-channel audio for the time required for dialogue detection (for example, about 0.5 seconds). It is preferable to provide it. In the case where multi-channel audio is accompanied by video, it is more preferable to apply a similar delay to the video signal. In the dialogue extraction circuit 11, a delay circuit TD is provided between the output of the band pass filter (BPF) 11a and the multiplier 11c. Since synchronization between the multi-channel sounds is maintained, the listener 2 can reproduce a good multi-channel sound field.

  The speech removal circuit 16 multiplies the speech extraction signal Cd0 by a coefficient (-1) and inverts the phase thereof, and subtracts the speech extraction signal Cd0 from the forward central signal C to obtain a speech removal forward central signal Cf0. It includes an adder 16b to be generated, a multiplier 16c that multiplies the speech removal front center signal Cf0 by a coefficient kc1, and further includes a multiplier 16d that multiplies the speech removal front center signal Cf0 by a coefficient kc2. In the present embodiment, the coefficient kc1 is set to 0, the speech removal front center signal Cf0 is not input to the adder 15, and the coefficient kc2 is set to 1 and the speech removal front center signal Cf0 is output as the output signal Cv.

  The DSP 10 of the multi-channel audio reproduction device 3 inputs the surround left signal SL and the surround right signal SR to the crosstalk giving circuit 17 of the DSP 10 as in the case of the previous embodiment. The outputs vs1 and vs2 of the lattice filter of the crosstalk applying circuit 17 are input to an adder 18 through multipliers 19a and 19b and a delay circuit TD, respectively. However, in this embodiment, the surround left signal SL and the surround right signal SR are branched before being input to the crosstalk giving circuit 17, and are respectively multiplied by multipliers 19c and 19d and a delay circuit TD. And then output as output signals Sl and Sr. Therefore, by setting the coefficient ks1 of the multipliers 19a and 19b to 0, and setting the coefficient ks2 of the multipliers 19c and 19d to 1, the surround left signal SL and the surround right signal SR are processed by the crosstalk applying process and the crosstalk. Without being subjected to virtual surround processing including talk cancel processing, the sound is output as it is to the surround left speaker SLsp and the surround right speaker SRsp.

  Since the output signal Cv is amplified and reproduced from the front center speaker Csp, the sound of the front center signal C excluding the speech sound, that is, the sound that should be localized in front of the listener 2 is the sound of the listener 2. Played from the front. The output signals Sl and Sr are output from the surround left speaker SLsp and the surround right speaker SRsp, and surround sound is reproduced from both sides of the listener 2. Therefore, the multi-channel sound reproducing apparatus 1 allows the listener 2 to perform the surround sound field reproduced from the front speaker and the surround speaker as shown in FIGS. 3A to 3D shown in the previous embodiment. Further, since only the emphasized speech is felt to be reproduced from the vicinity of the pinna of both ears of the listener 2, while the speech other than the speech is localized around the listener 2, the listener 2 It is possible to improve the intelligibility regarding speech with different feelings of localization.

  In the present embodiment, since the crosstalk cancellation process is performed only for the speech sound extracted based on the speech sound envelope signal, the addition of the signal can be minimized, so that the reduction of the dynamic range due to the addition of the signal can be suppressed. Of course, also in the present embodiment, the speech removal front center signal Cf0 may be reproduced from the front left speaker Lsp and the front right speaker Rsp at the same homologous level as in the previous embodiment. Alternatively, the surround left signal SL and the surround right signal SR may be subjected to virtual surround processing and output from the front left speaker Lsp and the front right speaker Rsp.

  In addition, when the multi-channel audio output from the decoding circuit 11 of the DSP 10 is a so-called 7.1 channel, the multi-channel audio reproducing device 1 or 3 further includes the surround back left signal and the surround back right signal that are included. It is also possible to provide a crosstalk application circuit that applies interaural crosstalk based on the transfer function of the virtual arrival angle, and input the output to the crosstalk cancellation circuit.

  Further, the selection input circuit 13 of the multi-channel sound reproducing apparatus 1 or 3 of the present embodiment includes multipliers 13a and 13b that multiply the speech emphasis signal Ce0 by coefficients ke1 and ke2, respectively, and the coefficients ke1 and ke2 are In addition to the values of 1, 0, and −1, other combinations and coefficients that vary with time may be used. Since the combination of the coefficients ke1 and ke2 fluctuates with time, the emphasized speech is reproduced from the vicinity of the pinna of both ears of the listener 2, and does not move. As a result, the speech can be clarified.

  The multi-channel audio reproduction apparatus of the present invention is not limited to the above-described embodiment, and is also suitable for an AV receiver that transmits a multi-channel audio signal format such as MPEG-2 / AAC and a multi-channel audio signal reproduction apparatus such as a disk reproduction apparatus.

It is a figure explaining the multichannel audio | voice reproduction apparatus 1 by preferable embodiment of this invention. Example 1 It is a figure explaining arrangement | positioning of the front speaker in the surround sound field which the multichannel audio | voice reproducing | regenerating apparatus 1 reproduces, and the listener's 2 head-related transfer function HRTF. Example 1 It is a figure explaining the localization feeling of the speech sound which the listener 2 obtains with the multichannel audio | voice reproduction apparatus 1. FIG. (Examples 1 and 2) It is a figure explaining the multichannel audio | voice reproduction apparatus 3 by preferable embodiment of this invention. (Example 2) It is a figure explaining arrangement | positioning of the front speaker in the surround sound field which the multichannel audio | voice reproduction | regeneration apparatus 3 reproduces, and the listener's 2 head-related transfer function HRTF. (Example 2) It is a figure explaining the dialog detection circuit 11b of the multichannel audio | voice reproduction apparatus 3. FIG. (Example 2)

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Multichannel audio | voice reproduction apparatus 2 Listener 3 Multichannel audio | voice reproduction apparatus 10
DESCRIPTION OF SYMBOLS 11 Dialog extraction circuit 12 Dialog emphasis circuit 13 Multiplier 14 Crosstalk cancellation circuit 15 Adder 16 Dialog removal circuit 17 Crosstalk provision circuit 18 Adder 19 Multiplier

Claims (7)

A speech extraction means for outputting a speech extraction signal that has passed through the speech voice band of the front center signal;
A line emphasizing unit that performs a line emphasis process for relatively increasing the level so that the upper limit of the line speech band of the line extraction signal is at a maximum level, and outputs the line emphasis signal;
The line emphasis signal is branched into a first line emphasis signal and a second line emphasis signal, and a crosstalk cancellation process for removing interaural crosstalk of the listener is performed on each of them to perform the first crosstalk cancellation signal and the first crosstalk cancellation signal. 2 crosstalk cancellation means for generating a crosstalk cancellation signal,
Playing back the first crosstalk cancellation signal from the left speaker and playing back the second crosstalk cancellation signal from the right speaker;
Multi-channel audio playback device. The dialogue extraction means includes dialogue detection means for detecting whether or not the dialogue speech is included in the front center signal,
When the line detection means controls the output level of the line extraction signal according to the level of the 1-4 Hz component level of the envelope signal of the line extraction signal and detects that the line speech is included, the line extraction signal The output level of the dialogue extraction signal is reduced when detecting that the speech is not included.
The multi-channel audio reproducing apparatus according to claim 1. Dialog removal means for generating a dialogue removal front center signal obtained by subtracting the dialogue extraction signal from the front center signal;
Adding means for adding the speech removal front center signal to the first and second crosstalk cancellation signals,
Playing the dialogue-removed front center signal and the first crosstalk cancellation signal from the left speaker, and playing the dialogue-removed front center signal and the second crosstalk cancellation signal from the right speaker,
The multi-channel audio reproduction apparatus according to claim 1 or 2. Line removal means for generating a line removal front center signal obtained by subtracting the line extraction signal from the front center signal,
Playing the speech-removed front center signal from a center speaker, playing the first crosstalk cancellation signal from the left speaker, and playing the second crosstalk cancellation signal from the right speaker;
The multi-channel audio reproduction apparatus according to claim 1 or 2. When the crosstalk canceling unit inputs only one of the first speech emphasis signal and the second speech emphasis signal to be subjected to the crosstalk cancellation processing, and when both are input at the same homology level, Including both a case of inputting both at the reverse homology level, and a selection input means for selecting and inputting either case,
When the selection input means is selected, the dialogue voice is emphasized to the listener, the dialogue voice is felt in the vicinity of either one of the listener's both ears, and the listener's head. Switching between the case of feeling the speech inside and the case of feeling the speech on the back of the listener,
The multi-channel audio reproducing apparatus according to any one of claims 1 to 4. The multi-channel audio reproduction apparatus according to any one of claims 1 to 5, further receiving a front left signal and a front right signal among the multi-channel audio.
A multi-channel audio reproducing device comprising: an adding means for adding the front left signal to the first crosstalk cancellation signal and adding the front right signal to the second crosstalk cancellation signal. The multi-channel audio reproducing apparatus according to claim 1, further receiving a surround left signal and a surround right signal among the multi-channel audio.
A cross for applying a crosstalk applying process for applying a binaural crosstalk based on a transfer function of a virtual arrival angle to the surround left signal and the surround right signal to generate a first crosstalk applying signal and a second crosstalk applying signal. Talk giving means,
A multi-channel audio reproduction apparatus comprising: adding means for adding the first crosstalk giving signal to the first speech emphasizing signal and adding the second crosstalk giving signal to the second speech emphasizing signal.

Images