JPH09247788A - Sound processing unit and conference sound system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To compensate deterioration in voice articulation in a reverberation sound field. SOLUTION: A digital filter 14 is interposed to a sound reproduction system of a tape recorder and a spatial transfer function corresponding to plural relative positions between a sound source and a microphone at recording is measured in advance and plural reciprocal filter coefficients Ga-Gn corresponding to the spatial transmission function are stored in a ROM 21. When the user operates a selection dial 22 to activate an address generating circuit 24 via a rotary encoder 23, a proper reciprocal filter coefficient is fed to a digital filter 14 from the ROM 21 and a reciprocal filter to a required characteristic is formed and the articulation of the sound reproduced from the speaker 17 is improved to allow the listener to hear sound easily.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice processing device and a conference voice system suitable for an environment with a lot of reverberant sounds.

[0002]

As is well known, in a general room,
The energy of the sound emitted from the sound source is accumulated in the room due to the reflection on the wall surface, and when the emission from the sound source is stopped, the energy is reduced almost logarithmically. In this way, the sound remains in the room until the emission from the sound source is stopped is called reverberation, and its magnitude greatly affects the acoustic effect in the room.

[0003]

By the way, in the reverberant sound field, the deterioration of the intelligibility of the voice may be caused by, for example, the Electroacoustic Research Society of Japan Material EA87-42 (August 1987).
, Etc.

Incidentally, in this document, in the reverberant sound field,
STI (Speach Transmiss
ion index) is expressed as an objective physical quantity. This STI is an envelope transfer function MTF (Modulation Tansf) that can be calculated based on the impulse response of the system.
ar Function).

As described above, since the intelligibility of a voice is deteriorated in a sound field with a lot of reverberation, when recording or picking up a sound in such a sound field, there is a problem that the target voice is difficult to hear. there were. In particular, when there are a plurality of sound sources (or speakers), the respective voices cannot be separated and are buried in reverberant sounds / reflected sounds, which makes it more difficult to hear the target voice.

In view of the above point, an object of the present invention is to provide a voice processing device and a conference voice system capable of compensating for the deterioration of the voice intelligibility in a reverberant sound field.

[0007]

In order to solve the above problems, a first aspect of the present invention is a voice processing device for processing a voice signal obtained from a sound source in a predetermined space via a sound collecting means. The inverse filter forming means for forming an inverse filter relating to the spatial transfer function between the sound source and the sound collecting means is provided, and the audio signal is processed through the inverse filter.

The conference audio system according to the second aspect of the present invention is a conference audio system for processing audio signals obtained from a plurality of attendees in a conference room via the audio collecting means. , An inverse filter formation for selectively forming a plurality of inverse filters respectively corresponding to the plurality of spatial transfer functions while installing the plurality of spatial transfer functions between the sound collecting means and the attendee different from each other. Have means,
The voice signal of any attendee is processed through an inverse filter corresponding to this attendee.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment in which a voice processing apparatus according to the present invention is applied to a magnetic recording / reproducing apparatus will be described below with reference to FIGS.

[Structure of Embodiment] FIG. 1 shows the structure of an embodiment of the present invention.

In FIG. 1, an analog audio signal taken out from a reproducing magnetic head 11 of an audio reproducing system 10 of a tape recorder is supplied to an AD converter 13 through a reproducing / amplifying circuit 12 and converted into a digital audio signal. Are supplied to the digital filter 14.

On the other hand, the ROM 21 stores a group of inverse filter coefficients Ga to Gn, a certain coefficient read from the ROM 21 is supplied to the digital filter 14, and an inverse filtering process as described later is performed. Done.

The digital audio signal output from the digital filter 14 is supplied to the D / A converter 15 to be converted back into an analog audio signal, supplied to the speaker 17 through the drive amplification circuit 16, and reproduced from the speaker 17. Sound is emitted.

Further, the selection dial 22 is coupled to the rotary encoder 23, and the output of the encoder 23 is supplied to the address generating circuit 24.
From the above, data of an appropriate address corresponding to the operation of the dial 22 by the user is supplied to the ROM 21, and one of the inverse filter coefficients Ga to Gn of 1 group stored in the addresses 21a to 21n of the ROM 21. Are selectively read out.

[Inverse Filtering Process of Embodiment] Next, the inverse filtering process of the embodiment of FIG. 1 will be described with reference to FIGS. 2 and 3.

In this embodiment, for example, in the case where an interview is conducted in a conference room, the room in which the recorder is used,
Assuming the recording situation such as the position of the sound source and the recorder,
The spatial transfer functions of the sound source and the recorder are measured in advance for a plurality of recording situations. This space transfer function also includes reverberation characteristics of the room itself.

When the required spatial transfer function as described above is, for example, an impulse response as shown in FIG. 2A, its inverse filter can be obtained by a least square method algorithm such as Levinson method. , For example, as shown in FIG. 2B.

In this embodiment, the inverse filters of the respective spatial transfer functions are obtained off-line for the plurality of recording situations as described above, and the coefficients Ga to Gn thereof are stored in the ROM.
It is stored in 21.

Then, when reproducing the magnetic tape MT recorded in the same manner as usual, the user operates the selection dial 22 while listening to the sound output from the speaker 17 to reverse the contents in the ROM 21. Filter coefficient G
The values a to Gn are sequentially read and supplied to the digital filter 14.

In the digital filter 14, the inverse filter calculation is performed in real time so that the user can improve the sound quality of the reproduced sound of the speaker 17 most, that is, the user can hear the sound with less reverberation. , An appropriate inverse filter coefficient is selected.

For example, as shown in FIG. 3, a person Pa who is speaking at a position near the video camera 1 of the reporter R,
In a recording situation in which there is a person Pd speaking at a position far from the video camera 1, the voice of the person Pd at the far position is likely to be masked by the voice of the person Pa at the near position, and further, the reverberation of the room is added, In normal reproduction, it becomes difficult to hear the sound of the person Pd located far from the video camera 1.

In this embodiment, the inverse filter coefficients of the spatial transfer functions Ha, Hd between the persons Pa, Pd and the video camera 1 in the recording situation as shown in FIG. It is stored in advance in the ROM 21 of the audio reproduction system 10.

The purpose of recording is the sound of the person Pd far from the video camera 1, and its spatial transfer function Hd.
Is represented by an impulse response as shown in FIG. 2A, the selection dial 22 is operated during reproduction,
Corresponding to the spatial transfer function Hd during recording, for example, FIG. 2B
The coefficients of the inverse filter as shown in are selected. The result of filtering (convolution) by the digital filter 14 based on this coefficient is, for example, as shown in FIG. 2C, and the original impulse can be reproduced although a time delay occurs.

In other words, the reverberation is added by the impulse response characteristic of the recording space, and the audio signal to which the frequency characteristic peculiar to the recording space is added is subjected to inverse filtering processing to obtain the added reverberation. , The frequency characteristic peculiar to the recording space can be removed, and the original audio signal with a flat frequency characteristic can be restored.
As a result, the deterioration of the intelligibility is compensated, and the reproduced sound becomes easy to hear.

Even if the actual transfer characteristic of the sound field and the inverse filter coefficient in the ROM are not the inverse characteristics in the strict sense, the reverberation and the frequency characteristics can be improved.
Further, in the above-described embodiment, the case of analog recording has been described, but in the case of digital recording, a required interface is used instead of the AD converter.

[Arrangement of Another Embodiment] Next, an embodiment of a conference audio system according to the present invention will be described with reference to FIGS. 4 and 5.

As shown in FIG. 4, in this audio system, it is assumed that a single microphone 31 is used commonly to all the participants Ma to Mn of the conference. The microphone 31 is omnidirectional in the horizontal plane. Also, FIG.
As shown in FIG.
a to 37n and selection keys Kaa to Kan, ..., Kna to Knn
And can be used.

The embodiment of the conference audio system according to the present invention is electrically constructed as shown in FIG. That is, in FIG. 5, the analog audio signal from the microphone 31 passes through the distribution / amplification circuit 32, and each participant Ma.
A plurality of AD converters 33 corresponding to Mn (see FIG. 4)
a to 33n are commonly supplied, converted into digital audio signals, and supplied to digital filters 34a to 34n. The digital audio signals output from the digital filters 34a to 34n correspond to the corresponding DA converters 35a to 35a.
n to be converted into analog audio signals, and transmitted through the drive amplifier circuits 36a to 36n to the headphones 37a to 37n.
n, and the sound picked up by the microphone 31 is reproduced from each of the headphones 37a to 37n.

The inverse filter coefficients Ga to Gn of the first group are RO
The selection keys Kaa to K stored in the predetermined addresses 41a to 41n of the M41 and corresponding to the participants Ma to Mn in the ROM 41 through the key interface 42.
An, ..., Kna to Knn are connected. One of the inverse filter coefficients Ga to Gn of the ROM 41 is selectively read according to the operation of the key Kia to Kin of any participant Mi,
The digital signal is supplied to the digital filter 34i, and a desired inverse filtering process is performed as in the above-described embodiment.

[Inverse Filtering Process According to Another Embodiment] Next, an inverse filtering process according to another embodiment of the present invention will be described with reference to FIG.

Also in this embodiment, the microphones 31 from the seats of the participants Ma to Mn are the same as in the above-described embodiments.
The space transfer functions up to are measured in advance, and the inverse filter coefficients Ga to Gn of each transfer function are stored in the ROM 41.

In this embodiment, the position of the microphone 31 is set so that the spatial transfer functions between the participants Ma to Mn are different from each other. It is also possible to use a plurality of microphones so as to satisfy this setting condition.

It is assumed that any participant Mi needs the voice of the participant Ma rather than the voice of the participant Mc.

In this case, the participant Mi selects the selection key Kia ...
By operating the key Kia corresponding to the participant Ma among Kin, the inverse filter coefficient Ga of the spatial transfer function Ha from the participant Ma to the microphone 31 is read from the ROM 41 and supplied to the digital filter 34i. ,
In the digital filter 34i, the inverse filter operation is performed in real time, and the voice signal of the participant Ma is inversely filtered as in the above-described embodiment.

As a result, the reverberation and the frequency characteristic, which are peculiar to the conference room, are removed from the voice signal of the participant Ma picked up by the microphone 31, and the headphone 37i participates with a flat frequency characteristic. The original voice of the person Ma can be obtained.

As described above, in the voice of the participant Ma, the deterioration of the intelligibility is compensated as a result of the inverse filtering process, the reproduced sound is easily heard, and the voice of the participant Mc which is not subjected to the inverse filtering process. In contrast, they can be separated auditorily.

As described above, when the participant Mi needs the voice of the participant Ma, when another participant, for example, Mn, needs the voice of the participant Mc, the participant Mn By operating the key Knc (not shown) corresponding to the participant Mc among the selection keys Kn to Kn, the spatial transfer function Hc from the participant Mc to the microphone 31 is read from the ROM 41 in the same manner as described above. Issued coefficient G
Based on c, the voice of the participant Mc, which has been inversely filtered and compensated for the deterioration of the intelligibility to make it easier to hear, is reproduced from the headphones 37n. Also in this case, the voice of the participant Ma that has not been subjected to the inverse filtering is
Aurally separable.

[Other Embodiments] In the embodiment shown in FIG. 1, the case where the present invention is applied to a reproducing system for magnetic recording has been described. However, the present invention is applicable to a recording system such as magnetic recording. The same can be applied, and when recording, the target sound can be set to remove the reverberation and frequency characteristics inherent in the recording space by performing inverse filtering processing. In this case, a monitor amplification circuit and headphones are required in addition to the recording amplification circuit.

In addition to the above-described embodiment, the present invention can be applied to a hearing aid and a telephone, and can also be applied to a voice input section of a voice recognition device.

Further, in the above-described embodiment, the digital filter is used for the inverse filtering, but an appropriate analog filter can be used instead.

Further, in the above-described embodiment, the transfer function of the recording or sound collecting space is measured in advance and the required inverse filter calculation is performed. However, the measurement of the spatial transfer function and the inverse filter calculation are performed. An online filter may be used to automatically select the optimum inverse filter for the situation of recording or sound pickup.

[0042]

As described above, according to the present invention, it is possible to remove reverberation and frequency characteristics peculiar to the sound collecting space from the sound signal picked up from the sound source.
Deterioration of the clarity of the reproduced voice is compensated for, making it easier to hear.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment in which a voice processing device according to the present invention is applied to a magnetic recording / reproducing device.

FIG. 2 is a waveform chart for explaining the operation of the embodiment of the present invention.

FIG. 3 is a conceptual diagram for explaining the operation of the embodiment of FIG.

FIG. 4 is a plan view for explaining the operation of the embodiment of the conference room audio system according to the present invention.

FIG. 5 is a block diagram showing a configuration of another embodiment of the present invention.

[Explanation of Codes] 10 ... Audio reproduction system, 11 ... Reproduction magnetic head, 14 ... Digital filter, 17 ... Speaker, 21 ... ROM (for storing inverse filter coefficient), 22 ... Selection dial, 23 ... Rotary encoder, 24 ... Address generation circuit, 30 ... Audio system (for conference room), 31 ... Microphone, 34a ...
34n ... Digital filter, 37a-37n ... Headphones, 41 ... ROM (for storing inverse filter coefficient), 42 ... Key interface, Ga-Gn ... Inverse filter coefficient, K
aa-Knn ... Selection key

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Takeshi Hara, 6-735 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation (72) Takayuki Mizuuchi 6-7-35, Kita-Shinagawa, Shinagawa-ku, Tokyo No. Sony Corporation (72) Inventor Kaoru Gyotoku 6-735 Kitashinagawa, Shinagawa-ku, Tokyo Sony Corporation (72) Ikue Akiba 6-735 Kitagawa-Shinagawa, Shinagawa-ku, Tokyo -Inside the corporation

Claims (6)

[Claims] 1. A voice processing device for processing a voice signal obtained from a sound source in a predetermined space through a sound collecting means, wherein the voice processing apparatus forms an inverse filter relating to a spatial transfer function between the sound source and the sound collecting means. An audio processing device having an inverse filter forming means for processing the audio signal through the inverse filter. 2. The inverse filter forming means selectively forms a plurality of the inverse filters in correspondence with a plurality of relative positions of the sound source and the sound collecting means in the predetermined space, respectively. The voice processing device described. 3. The audio processing apparatus according to claim 1, wherein the audio signal obtained through the sound collecting means is once recorded on a medium, reproduced from the medium, and processed by the inverse filter. 4. The inverse filter forming means includes digital filter means and memory means, and the memory means stores inverse filter coefficients corresponding to relative positions of the sound source and the sound collecting means in the predetermined space. The voice processing apparatus according to claim 1, wherein 5. The inverse filter forming means includes digital filter means, memory means and selecting means, and the memory means corresponds to a plurality of relative positions of the sound source and the sound collecting means in the predetermined space. The voice processing apparatus according to claim 1, wherein a plurality of inverse filter coefficients stored in the predetermined area of are selected by the selecting means. 6. A conference audio system for processing an audio signal obtained from a plurality of attendees in a conference room via a sound collecting means, wherein the sound collecting means is provided between the sound collecting means and the attendees. Are installed so that the plurality of space transfer functions between them are different from each other, and there is an inverse filter forming means for selectively forming a plurality of inverse filters corresponding to the plurality of space transfer functions. A conference audio system adapted to process an audio signal through the inverse filter corresponding to the attendee.