JPH0355771B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an audio device having a delay time detection function for detecting an optimal delay time of reflected sound that provides more preferable auditory characteristics.

There is a close relationship between the human auditory psychology or sense of hearing and the physical characteristics such as the reverberation time of a room or the acoustic characteristics of the sound source itself.

A well-known human auditory model is Licklider's autocorrelation model of the auditory nervous system, which is said to demonstrate the close relationship between the sense of hearing and the sound source. .

FIG. 1 is a model of the auditory nervous system using a retrick rider. In Figure 1, 1 is the direct nerve, 2
is the delay nervous system, 3 is the input, 4 is the output, and 5 is the synapse. An impulse synchronized with the envelope period of the original signal after some frequency analysis in the cochlea comes in from input 3. When the first impulse enters, it travels through direct nerve 1,
The delayed nervous system 2 that has acquired the synapse 5 is also transmitted. When the second impulse enters, its direct nerve 1
The upper signal catches up with the delayed first impulse, and due to the simultaneous occurrence of this impulse, output 4
is excited and a time-place transformation is performed. In other words, the human auditory nervous system has a self-interpersonal function.

Recent research on hearing has revealed that reflected sounds that arrive later than direct sounds, and subsequent reflected sounds, have an important effect on hearing.

J.Acoust.Soc.Am.Vol.62.No.6, December
1977 (P1436-1441) “Subjective Preference in
relation to objective parameters of music
"Fields with a Single Echo" uses music and speech to evaluate a synthesized sound field consisting of direct sound and a single reflected sound when reproduced by a speaker, using a pref (human auditory comfort) scale. .
According to this result, when the normalized autocorrelation function _p (τ) of the sound source signal is determined and the level of the reflected sound is varied over ±6 dB of the direct sound, the optimal delay time of the reflected sound is determined. It has become clear that | _p (τ) | corresponds to a time corresponding to 1/10 of the level A ₁ of the first reflected sound. Figure 2 shows this | _p (τ)
The horizontal axis represents the time (γd) corresponding to 1/10 of the level A ₁ of the first reflected sound, and the vertical axis represents the delay time τm of the single reflected sound at which the preference is maximum. The range shown in the figure shows the delay time when it is 0.1 lower than the maximum value of the preference, and the symbol ○ in the figure indicates A ₁ = 6 dB, and the symbol ● indicates A ₁ =
0dB, □ indicates A ₁ =-6dB. In particular, |
If the time when _p (τ) | becomes 0.1 times _p | is called τe (0.1), when A ₁ = 0 dB, it can be expressed as τd = τe (0.1). From Figure 2, τd is the delay time τm of a single reflected sound at which the preference is maximum.
It can be seen that there is good agreement.

In addition, the Acoustical Society of Japan Proceedings May 1980 (P677-678) “Subjective Preferences of Hearing Impaired Children”
and Audiology Japan Vol.23, No.5, 1980
(P301-302) In “Subjective evaluation of sounds with echoes by children with hearing loss,” a similar experiment was conducted for monaural listening using earphones, and the preference of a single reflected sound in this case was investigated. The maximum value is the time when | _p (τ)| becomes 0.25 times _p |, τe
It is clear that it can be obtained at (0.25). In Figure 3, the horizontal axis is the delay time △ of a single reflected sound.
t ₁ , the vertical axis represents the normalized preference, where a is for speech and b is for music. In the figure, three graphs for those with normal hearing, those with a hearing loss (1 KHz) of 70 to 95 dB, and those with a hearing loss (1 KHz) of 100 dB are represented by ○, ●, and ▲, respectively.
As is clear from this graph, the preference reaches its maximum value when the delay time of a single reflected sound is τe (0.25), not only for people with normal hearing but also for those with a hearing loss of 95 dB or less. In other words, this result shows that if this is taken into consideration when designing a hearing aid, a hearing aid with characteristics that are easier to hear can be obtained.

Furthermore, it has been reported that the autocorrelation of the sound source signal is closely related to the optimal reverberation time. Figure 4 shows the measurement results. The horizontal axis shows the aforementioned τe (0.1), and the vertical axis shows the preferred median value of reverberation time [Tsub]d. The reverberation time here is not expressed as the time for the direct sound to attenuate by 60 dB, but as the time for the signal in the reverberation section to attenuate by 60 dB. In the figure, A, B, and E represent music, and S represents speech, which can be approximately approximated by the function [Tsub]d≈(23±10)τe(0.1).

In view of the above-mentioned situation, it is an object of the present invention to provide an acoustic device that applies a detection function to determine the delay time or reverberation time of reflected sound that is preferable to human auditory sense from the autocorrelation function of the sound source.

Hereinafter, one embodiment of the present invention will be described with reference to the drawings. In Fig. 5a, 21 is an input terminal;
2 to 24 are delay circuits, 25 to 28 are multipliers, 29
- 32 are integration circuits, 33 to 36 are absolute value conversion circuits,
37 is an attenuator, 38 to 40 are comparison circuits, and 41 is an encoder.

Here, the function and operation will be explained. Input terminal 1
When an acoustic signal is applied to , it is transmitted to multipliers 25 to 28 and delay circuit 22 . Each delay circuit 22
~24' provides a plurality of different delay outputs D ₁ , D ₂ ,...
...DN-1 is obtained, and the input signals are multiplied by each other or by each delayed output in multipliers 25-28. Thereafter, the values are integrated in the respective integration circuits 29-32, and the absolute values are determined by the absolute value converting circuits 33-36.

It goes without saying that the integrating circuit generally needs to be initialized before starting the integrating operation, and in some cases a low-pass filter can be used instead. Outputs of the absolute value conversion circuits 33 to 36 φ ₀ , φ ₁ ,
..., φ _N-1 corresponds to the absolute value of the autocorrelation function of the input acoustic signal. FIG. 5b shows φ ₀ , φ ₁ , ..., φ _N-1 when the signal is obtained by removing high frequency components from the input using a low-pass filter,
This level is sufficient when configuring as a simple form.
φ ₀ is the maximum value, and as φ ₁ , φ _{2 .} . . , the value generally decreases.

Absolute output of signal without time delay φ ₀
is attenuated to 1/10 or 1/4 by an attenuator 37, and then output to other absolute value outputs and comparison circuits 38 to 40.
compared by. The comparator circuits 38 to 40 provide the attenuator output of the signal without time delay.
An absolute output smaller than Ref is selected, and a comparison output related to the delayed output with the smallest delay time is selectively encoded by the encoder 41. Note that the comparison output related to the longest delayed output, which is larger than Ref, may be encoded.
That is, the delayed output corresponding to the addition means having the addition output closest to the attenuation output Ref may be selected.

In addition, the damping ratio 0.1 or 0.25 mentioned above is 0.1±0.3 or
Almost the same effect can be achieved even within the range of 0.25±0.08.

FIG. 6 shows a first application example constructed using the delay time detector according to the present invention. This is an optimal reflected sound addition device that can be applied to acoustic devices (including hearing aids) to create more desirable and easier-to-hear sounds. In FIG. 6, 21 to 41 constitute the delay time detector of FIG.
Perform a similar action. τe(0.1) or τe(0.25) of the input signal required to add the optimal single reflection
An output corresponding to is obtained from encoder 41 and stored in latch 42 for holding for a not too short time. The optimum delay time for a single reflected sound is configured to latch at intervals of not too short a time, since it is not necessary to do it rapidly and frequently.
The attenuator 44 is designed to gently attenuate the delayed signal at the time of switching the latch so as not to give the user a sense of noise. This latch signal generator is not shown. Depending on the output of the latch 42, an appropriate one is selected from among the delayed outputs D ₁ , _{D 2 ,} . 45 to obtain an output 46.

With such a configuration, it is possible to realize an audio device that has a large preference (aural comfort) and is easy to listen to.

FIG. 7 shows another application constructed using the delay time detector according to the invention. This is an optimal reverberation adding device that can be applied to a sound reproduction device to create a more desirable and easy-to-hear sound. In FIG. 7, numerals 21 to 41 constitute the delay time detectors in FIGS. 5 and 6, and perform similar operations. The code converter 41 receives the comparison signal S ₁ -S _N-1, which is a signal necessary for setting the optimum reverberation time.
obtained by. This signal is latched at an appropriate timing as in the case of FIG.
This latched signal is a reverberation time setting signal and is applied to the reverberation circuit 47. The reverberation circuit 47 receives the input signal supplied to the input terminal 21 and the reverberation time setting signal and generates a reverberation addition signal output 46.

With such a configuration, it is possible to realize an optimal reverberation adding device that has the greatest preference (aural comfort) and is easy to listen to.

As described above, according to the present invention, it is possible to construct a hearing aid that is more preferable, has high preference, and provides easy-to-hear sounds, and various other acoustic devices.

In the second application example shown in FIG.
Although the details of the delay circuit 22 were not specified, any method that can set the reverberation time variably, such as a delay element such as a BBD or a mechanical delay element such as a spring, may be used.
It may be configured to use output signals such as .about.24.

In the above explanation, as a simplified form for determining the values of φ ₀ , φ _{1 ,} . The reason for this is φ ₀ , φ ₁ ,...
This is because consideration was given so that φ _N-1 could be obtained as a monotonically decreasing function as shown in FIG. 5b. Similarly, the fifth
A method for obtaining the function shown in Figure b is to provide a filter in front of the input 21 that allows only periodic components that are integral multiples of the delay time of the delay circuit to pass, and then process the filter using the circuit described above. One example is doing something. In this case, although it is somewhat complicated, high frequency components are also taken into consideration, and the degree of approximation also increases.

Furthermore, the block diagram of the correlator shown in FIG. 5 is shown in a formal manner, and it is not necessarily necessary to use individual circuits, but it may be performed by calculation processing using a microcomputer, etc. In D.
-A (digital to analog) converter to point P, A
- A D (analog-digital) converter is provided at point Q,
It goes without saying that the signal processing may be performed using digital data. Furthermore, although the method of configuring this correlator has been described as a very usual method, a configuration such as Schroeder's square integral type reverberator may also be used.

Next, an embodiment using this square integral type reverberator will be described with reference to FIG.

In FIG. 8, an audio signal is applied to the input terminal 21. Outputs of the integrating circuits 29 to 32 φ ₀ , φ ₁ ,...
...φ _N-2 and φ _N-1 are obtained as in the first embodiment. The integral outputs φ ₀ , φ ₁ , . . . φ _N-1 are squared by multipliers 51 to 54, respectively. The squared outputs are respectively added, that is, squared and integrated, by adders 55 to 57, and the output of adder 57 is attenuated by attenuator 37. This damping ratio is the square value of the value in the first embodiment of FIG. 5a, i.e.
It is preferable to use a value within the range of 0.01±0.003 or 0.0625±0.02. Thereafter, the comparison circuits 38 to 40 compare the output Ref of the attenuator 37, and the output that is smaller than Ref and closest to the adder 57 is selectively encoded by the encoder 41. It goes without saying that the output of the adder 56 is larger than the output of the adder 55 in terms of the circuit, and the output of the similar adder 57 is the largest. In the case of the second embodiment, the signal that serves as the reference for determining the optimal delay time to be determined is the signal from the integrating circuit 3, contrary to the first embodiment.
It should be noted that the output of the adder 57 corresponds to the signal of φ _N-1 obtained from the output of the adder 57.

Furthermore, a method for determining φ ₀ , φ ₁ , . . . φ _N-1 more precisely will be described. FIG. 9 shows an example of its configuration. In figures a and b, 61 is a delay circuit, 62 is a multiplier, 63 is an integration circuit, 64 is an absolute value converting circuit, 65 is a delay circuit 61, and a multiplier 6.
2. A circuit A including an integrating circuit 63 and an absolute value converting circuit 64, 66 is a maximum value detecting circuit. In Figure 9a, SIN is an input terminal, DIl is a delay input terminal,
Dl is a delay output terminal, and φl is a correlation output terminal. Correlation output φl by giving SIN and DIl
get. Using this circuit A, φ ₀ , φ ₁ , . . . φ _N-1 are determined by the configuration shown in FIG. 9b.
φ ₀ is obtained by integrating the square value of the input signal, and it goes without saying that the absolute value conversion circuit 33 is not necessarily required. Then circuit A ₁ has DI1, SIN
is given and we get φ ₁ . Similarly, below φ ₂ , φ ₃ ,...
For example, in this _embodiment , the maximum value detection circuit 66 selects the maximum value from among the ten φl values and sets it as _φ0 . Hereinafter, φ ₁ , φ ₂ , . . . φ _N-1 will be obtained using the same method. Normally, there is no need to calculate it; φ ₀ is the largest, so
The circuit for deriving φ ₁ to _{φ 9} and the maximum value detection circuit 66 for determining φ ₀ can be omitted. As mentioned here, when φ ₀ , φ ₁ , ... φ _N-1 are determined, even if the acoustic input signal contains high frequency components, their values are functions that decrease almost monotonically as shown in Figure 5b. The optimal delay time can be obtained with high accuracy. In this case, the optimal delay time is the delay time corresponding to the output of the maximum value detection circuit closest to Ref.
A predetermined one is selected.

FIG. 10 shows a further improvement of the embodiment shown in FIG. In the figure, 70 is a comparator and 71 is an AND
It is a gate. In this case, unlike the embodiment shown in FIG. 9, φ ₁ , φ ₂ , ...φ _N-1 are not determined,
The inputs S ₁ , S ₂ , . . . _SN-1 of the encoder 41 can be obtained. The values of φ ₀ , φ ₂ , . . . φ ₁₀ (N-1) may also be determined in substantially the same manner. Signal φ ₀ is attenuator 3
7 to create Ref, φ ₁ , φ ₂ ,...
φ ₁₀ (N−1) is compared by the comparator 70. The comparator 70 has φ ₁ , φ ₂ , ..., φ ₁₀ (N-1) than Ref.
is smaller, the output becomes a high level (hereinafter abbreviated as "H"). Signals S ₁ , S ₂ , ...S _N-1 are φ ₁ ~ φ ₁₀ , φ ₁₁ ~, respectively.
φ ₂₀ , ..., φ ₁₀ (N-2) + 1 to φ ₁₀ (N-1)
It is the output of the AND gate 71, and the inputs φ ₁ to _{φ 10} , φ ₁₁ to φ ₂₀ , φ ₁₀ (N-2)+1 to φ ₁₀ (
N-
1) becomes "H" when all of them become smaller than Ref. encoder 41
AND with output “H”
AND gate 71 or output “L” to which the delay circuit with the shortest delay time belongs among the gates 71
AND gate 7 with (meaning low level)
It outputs the delay time of a predetermined delay circuit corresponding to the AND gate 71 to which the delay circuit with the longest delay time among the delay circuits belongs.

With this configuration, the AND gate 71 is required, but the maximum value detection circuit 66 is not required.
This is advantageous because it simplifies the circuit configuration.

In both the practical examples shown in Figs. 9 and 10, one φl (l=0, N−
1), Sm (m=1, N-1) is determined, but it goes without saying that the number is not necessarily 10 and may be any other number.

[Brief explanation of drawings]

Figure 1 is a diagram showing a model of the human auditory system based on a Ritsuklider, Figure 2 is a diagram showing the relationship between τd and the delay time of a single reflected sound at which the preference is maximum, and Figures 3 a and b are diagrams showing the Delay time of one reflected sound △t ₁
Figure 4 shows the relationship between τe (0.1) and the preferred median reverberation time [Tsub] d, Figure 5 a and b show the relationship between Information τe for determining time and reverberation time
A block diagram and an operation explanatory diagram showing one embodiment of the delay time detector used in the present invention for determining τd, FIG. A block diagram showing the first reflected sound adding device, FIG. 7 is a block diagram showing an optimal reverberation adding device as a second application example using the delay time detector according to the present invention, and FIG. 8 is a block diagram showing the optimal reverberation sound adding device used in the present invention. A block diagram showing a second embodiment of a delay time detector used in the present invention, a diagram showing a partial configuration of FIGS. 9a and 9b, and a block diagram showing a third embodiment of a delay time detector used in the present invention FIG. 10 is a block diagram showing a fourth embodiment of the delay time detector used in the present invention. 1... Direct nerve, 2... Delay nervous system, 3... Input, 4... Output, 5... Synapse, 21... Input terminal, 22-24... Delay circuit, 25-28...
Multiplier, 29-32...Integrator circuit, 33-36...
...Absolute value conversion circuit, 37...Attenuator, 38-40...
... Comparison circuit, 41 ... Encoder or code converter, 42 ... Latch, 43 ... Multiplexer, 44 ... Second attenuator, 45 ... Adder, 4
6... Output, 47... Reverberation circuit, 61... Delay circuit, 62... Multiplier, 63... Integrating circuit, 64...
... Absolute value converting circuit, 65 ... Circuit A, 66 ... Maximum value detection circuit, 70 ... Comparator, 71 ... AND gate. Note that the same reference numerals in the figures indicate the same or equivalent parts.

Claims (1)

[Claims] 1. An input terminal to which an acoustic input signal is applied, a delay means for delaying the input signal and generating a plurality of delayed outputs having different delay times, and connecting the input signals to each other or the input signal and the delayed output. A multiplication means that generates a plurality of multiplication outputs by multiplying each of them, an integrating means that integrates each of the above multiplication outputs and generates a plurality of integral outputs, and an absolute value of each of the above integral outputs to generate a plurality of absolute value outputs. Absolute value conversion means for generating, attenuation means for generating an attenuated output by attenuating the absolute value conversion output related to the multiplication output of the input signals among the absolute value conversion output by a predetermined attenuation ratio, and the above absolute value conversion output. Comparing means for generating comparison outputs by comparing the absolute value output that is not related to the attenuation output and the attenuation output, and the attenuation output that is related to the comparison output having a predetermined comparison result among the comparison outputs. An audio device comprising means for outputting a delayed signal of a delayed output corresponding to the absolute value converting means having an absolute value converting output closest to , and an adding means for adding the input signal and the delayed signal. 2. The acoustic device according to claim 1, wherein the delayed signal is added to the input signal after its delay time is multiplied by (23±10).