JPH086596A - Voice emphasis device - Google Patents

Abstract

PURPOSE:To provide a voice emphasis device which is applicable to a telephone set, a loudspeaker and other acoustic instruments or a voice coding and decoding device and improves the clarity of a voice without deteriorating naturalness and makes the voice easy to hear by an aged person. CONSTITUTION:The device is provided with a voice feature parameter fluctuation amount analysis means 2 which divides an inputted voice into short time blocks and analyzes the degree of timewise fluctuation of the voice feature parameters of the inputted voice every block, an emphasis strength control means 3 which controls the spectrum emphasis strength in response to the voice feature parameter fluctuation amount information outputted by the means 2, a spectrum emphasis means 6 which executes the emphasis based on the means 3, a detection means which detects the voice that is deteriorated by the voice emphasis and noise and an emphasis coefficient change means 5 which suppresses or inhibits the execution of an emphasis based on more than one control signal outputted by the detection means.

Description

Detailed Description of the Invention

[0001]

INDUSTRIAL APPLICABILITY The present invention is used in telephones, loudspeakers, acoustic devices, and other voice coding / decoding devices.
The present invention relates to a voice emphasizing device that improves the clarity of voice while maintaining naturalness and makes it easier to hear the voice.

[0002]

2. Description of the Related Art Conventionally, as a method for improving the clarity of voice, there is, for example, JP-A-58-184200. In the method of improving the clarity of speech described in Japanese Patent Laid-Open No. 58-184200, the error in the perception of consonants caused by the fact that the power of consonants is considerably smaller than that of vowels in interactive speech is reduced. In order to do so, the consonant part is detected and the power of that part is increased to emphasize the part.

The operation of a conventional configuration example of a method for improving the clarity of voice will be described below.
The input speech is divided into short time periods, and the short time spectral morphology of the input speech is evaluated by inputting it to the spectrum morphology evaluation device for each short time period. Then, the spectrum form evaluation device generates one or more evaluation signals according to the evaluation contents of the spectrum form. The control logic device generates a control signal for controlling the spectrum morphology dynamic correction device according to the evaluation signal, and according to the control signal, each short time section of the input voice by the spectrum morphology dynamic correction device. The spectrum form is corrected for each and the output voice is output.

Further, as a method for converting the voice quality to improve the clarity of the voice, for example, Japanese Patent Laid-Open No. 01-93796.
No. In the method described in Japanese Patent Laid-Open No. 01-93796 for improving voice clarity by converting voice quality,
The emphasis is made by emphasizing the transition of the formant frequency, which is important as voice information.

The operation of a voice quality conversion method disclosed in Japanese Patent Laid-Open No. 01-93796 will be described below.
The input voice is divided into short time sections, and the voice analysis is performed by the analysis unit for each short time to distinguish between voiced sound and unvoiced sound.
When the input short time section is the voiced sound section, the linear prediction coefficient of the short time section is calculated to calculate the resonance frequency.
Next, in the formant frequency control unit, the formant frequency is obtained based on the resonance frequency obtained by the analysis unit, and the formant frequency and the bandwidth are changed so that the amount of change in the time trace of the formant becomes large. The spectrum control unit changes the spectrum envelope in accordance with the time variation of the formant frequency output by the formant frequency control unit and the bandwidth. When the process of voice quality conversion for a series of voiced sounds described above is completed, the unvoiced section and the unvoiced section are connected, the process proceeds to the next voiced section, and the finally synthesized output speech is output.

Further, as a method for improving the quality of synthesized speech by utilizing the dynamic characteristics of speech, reference 1, Ando et al.,
"Melk cepstrum voice considering dynamic features of voice" Technical Report of IEICE, SP93-65 (1)
993-10).

[0007] Document 1 proposes a method of using a two-dimensional mel cepstrum for enhancing dynamic features of speech. The processing of the voice quality improving method shown in Document 1 will be described below. The input speech is divided into short time sections, and the two-dimensional mel cepstrum is obtained using the current frame and the frames before and after the current frame. The dynamic feature region of the two-dimensional mel cepstrum is lifted by using a low-frequency weighting type weighting lifter in the time and frequency directions, and the low-frequency enhancement is performed by changing the shape of the spectrum envelope in the time direction and the frequency direction. Find a two-dimensional mel cepstrum. A single-pulse sound source is used as the sound source information, and the synthesized speech is synthesized by an MLSA (mel log spectrum approximation) filter and output. In addition, the explanation as a physical quantity of the two-dimensional mel cepstrum is
Since it is described in detail in Reference 1, it is omitted.

[0008]

Since the conventional speech emphasizing apparatus is constructed as described above, it has the following problems. In Japanese Patent Laid-Open No. 58-184200, the short-term spectrum of an input voice is evaluated, and the spectrum form is adaptively changed accordingly to emphasize the voice. In other words, since the emphasis is based on the "instantaneous value" of the voice feature parameter, it is possible to detect and enhance the consonant section, but the idea is to control the emphasis based on the "variation amount" of the voice feature parameter. However, there is a problem that it is not possible to emphasize the transitional part of the phoneme, which is important for the intelligibility of the voice.

According to Japanese Patent Laid-Open No. 01-93796, the amount of variation in the formant frequency is increased by emphasizing the formant time variation range, but the voice in the central section of the change is not emphasized and is important for the intelligibility of the voice. There was a problem that the transitional part of the phoneme could not be emphasized. In addition, the temporal locus of the formant frequency is modified so that its variation width is large on the frequency axis.Therefore, in order to increase the absolute value of the variation width, it is necessary to shift the position of the formant frequency even in the steady part. As a result, the formant frequency, which is important as the personality information of the voice, changes, and although the clarity of the part of the formant with a large temporal trajectory change is improved, the resulting output voice has a different voice quality and naturalness. However, there was a problem in that the personality was significantly impaired and the individuality was lacking. In addition, there is a problem that it is not possible to emphasize voices such as consonants whose temporal loci of formant peaks cannot be extracted.

In the synthetic speech quality improvement method shown in Document 1, the spectrum temporal frequency variation width is emphasized, but the speech in the center of the change is not emphasized, and the phoneme which is important for the intelligibility of the speech. There was a problem that the transitional part of could not be emphasized. Although the weighting coefficient of the lifter is changed according to the time frequency direction,
In this method, it is not possible to adopt a configuration in which the lifter is changed according to the phoneme shape of the input speech, and multiple enhancement methods are introduced to adaptively switch the methods according to the phoneme shape of the input speech. Also, it is not possible to operate multiple enhancement methods at the same time to achieve a synergistic effect.Therefore, adaptive enhancement according to the phoneme shape of the input speech, or speech enhancement using the synergistic effect of multiple enhancements There was a problem that I could not do it.

In any of the above methods, deterioration of the amplitude or distortion may occur due to the emphasis of the input voice, but the quality of the output voice deteriorates because no means for suppressing the occurrence of these is taken. There was a problem of doing.

The present invention has been made to solve the above problems, and an object of the present invention is to provide a voice emphasizing device which improves the clarity of the voice while maintaining the naturalness and makes the voice easy to hear.

[0013]

According to a first aspect of the present invention, there is provided a voice emphasizing device, which comprises a voice feature parameter variation amount analyzing means for analyzing a degree of temporal variation of a voice feature parameter of an input voice for each short time section, and a voice feature parameter variation amount analyzing means. Spectrum emphasis strength control means for controlling the spectrum emphasis strength in response to the voice feature parameter fluctuation amount information output from the feature parameter fluctuation amount analysis means, and the spectrum of the input voice based on the spectrum emphasis coefficient output from the spectrum emphasis strength control means And a spectrum enhancing means for enhancing.

A voice emphasizing apparatus according to a second aspect of the invention is a voice feature parameter variation amount analyzing means for analyzing a degree of time variation of a voice feature parameter of an input voice for each short time section, and a voice feature parameter variation amount analyzing means. And a pre-emphasis coefficient control means for controlling the pre-emphasis coefficient in response to the voice feature parameter fluctuation amount information output by the pre-emphasis means, and a pre-emphasis means for performing input voice pre-emphasis based on the pre-emphasis coefficient output by the pre-emphasis coefficient control means. Is provided.

A voice emphasizing apparatus according to a third aspect of the present invention is a voice feature parameter variation amount analyzing means for analyzing a degree of time variation of a voice feature parameter of input voice divided into short time intervals, and a voice feature parameter variation amount analysis. In response to the first speech feature parameter variation amount information outputted by the means, the spectrum emphasis strength control means for determining the spectrum emphasis strength and outputting the spectrum emphasis coefficient, and the spectrum emphasis coefficient outputted by the spectrum emphasis strength control means A pre-emphasis coefficient that determines a pre-emphasis coefficient and outputs a pre-emphasis coefficient in response to the spectrum emphasis means for enhancing the spectrum of the input voice and the second voice feature parameter fluctuation amount information output by the voice feature parameter fluctuation amount analysis means. The control means and the pre-emphasis coefficient output by the pre-emphasis coefficient control means A pre-emphasis unit for performing pre-emphasis of the input voice in accordance with the cis coefficient, is obtained as constructed from.

A speech emphasizing apparatus according to a fourth aspect of the present invention is the speech emphasizing apparatus according to the first aspect of the present invention, in which the input speech feature parameter divided into short time periods is input and the distortion is suppressed in the output speech. The present invention is provided with an emphasis incompatible voice detecting means for outputting a signal and an emphasis coefficient changing means for correcting the spectrum emphasis coefficient based on the emphasis suppression signal.

A voice emphasizing device according to a fifth aspect of the present invention is the voice emphasizing device according to the fourth aspect of the invention, wherein the emphatic nonconforming voice detecting means inputs the feature parameter information of the output voice in addition to the voice feature parameter information regarding the input voice. Thus, the emphasis suppression signal is controlled.

A speech emphasizing apparatus according to a sixth aspect of the present invention is the speech emphasizing apparatus according to the fourth aspect of the invention, wherein the emphasis coefficient changing means inputs the external control signal in addition to the emphasis suppressing signal to determine the spectrum emphasis coefficient. It is something that is done.

A speech emphasizing apparatus according to a seventh aspect of the present invention is the speech emphasizing apparatus according to the first aspect of the invention, wherein a spectral emphasis coefficient map defining the relationship between the speech feature parameter variation information and the spectral emphasis coefficient is used, A spectrum emphasis temporary coefficient determining means for calculating a spectrum emphasis coefficient is provided.

A speech emphasizing apparatus according to an eighth aspect of the present invention is the speech emphasizing apparatus according to the first aspect of the invention, in which a pre-emphasis coefficient map defining a relationship between speech feature parameter variation amount information and a pre-emphasis coefficient, And a pre-emphasis coefficient determining means for calculating an emphasis coefficient.

A speech emphasizing apparatus according to a ninth invention is the speech emphasizing apparatus according to any one of the first invention or the third invention to the seventh invention, in which a constant bias value is always added to the speech feature parameter variation amount information. However, the spectrum of the input voice is always emphasized regardless of whether or not the characteristic parameter of the input voice changes.

A voice emphasizing device according to a tenth aspect of the invention is the second aspect.
In the voice emphasizing device according to the present invention, the third aspect of the invention, or the eighth aspect of the invention, a constant bias value is constantly added to the voice feature parameter variation amount information, and the constant input voice characteristic is maintained regardless of whether the feature parameter of the input voice varies. It is designed to perform pre-emphasis.

A speech emphasizing device according to an eleventh aspect of the invention is the seventh aspect.
The speech enhancement apparatus according to the invention or the eighth invention is provided with a plurality of spectrum enhancement coefficient maps or a plurality of pre-emphasis coefficient maps.

[0024]

The speech emphasizing apparatus according to the first aspect of the present invention obtains a spectrum emphasis coefficient according to the variation amount of the speech feature parameter of the input speech, and adaptively emphasizes the input speech based on the coefficient value. A voice emphasizing device according to a second aspect of the present invention obtains a pre-emphasis coefficient according to a variation amount of a voice feature parameter of an input voice, and adaptively emphasizes the input voice based on the coefficient value. A speech emphasizing apparatus according to a third aspect of the present invention obtains a spectral emphasis coefficient and a pre-emphasis coefficient in response to a variation amount of a speech feature parameter of input speech, and adaptively emphasizes the input speech based on these coefficient values. To do. The speech emphasizing device according to the fourth aspect of the invention controls so as to suppress or inhibit the emphasizing of the input speech when an abnormal increase of the speech amplitude or noise mixing is expected due to the emphasizing process. A voice emphasizing device according to a fifth invention is
Control is performed so as to suppress or prohibit the enhancement of the input voice based on both the voice feature parameters of the input voice and the output voice. The voice emphasizing device according to the sixth aspect of the invention controls so that the emphasis of the input voice is suppressed or prohibited by an external input control signal in addition to the voice feature parameter of the input voice. Seventh
The speech emphasizing apparatus according to the invention of claim 1 derives a spectral emphasis coefficient from the smoothing process of the preceding and following audio frames based on the spectral emphasis temporary coefficient derived from the spectrum emphasis map to perform the emphasis process of the input voice. A speech emphasizing apparatus according to an eighth aspect of the present invention derives a pre-emphasis coefficient from a smoothing process of preceding and following speech frames based on a pre-emphasis provisional coefficient derived from a pre-emphasis coefficient map, and enhances an input speech. According to a ninth aspect of the present invention, there is provided a speech enhancement apparatus according to any one of the first aspect of the invention, the third aspect of the invention, or the seventh aspect of the invention, wherein a constant bias value is always added to the speech feature parameter variation amount information to input speech. The input speech spectrum is always emphasized regardless of whether or not there is a change in the characteristic parameter. First
The voice enhancement device according to the invention of No. 0 is the second invention or the third invention.
In the voice emphasizing device according to the invention or the eighth invention, a constant bias value is always added to the voice feature parameter variation amount information to always perform pre-emphasis of the input voice regardless of the variation of the feature parameter of the input voice. It is the one. A voice emphasizing device according to an eleventh invention is
The speech enhancement apparatus according to the seventh invention or the eighth invention is provided with a plurality of spectrum enhancement coefficient maps or pre-emphasis coefficient maps so as to take into consideration the voice type and environmental changes.

[0025]

【Example】

Example 1. A first embodiment of the present invention will be described below with reference to FIGS. 1 and 2. The "spectral enhancement" mentioned in the first invention is, for example, a formant peak of speech is extracted by using a linear predictive analysis method or a fast Fourier transform, and formant emphasis is performed to emphasize it, or band division is performed for each frequency band. This is equivalent to band-wise emphasis that is multiplied by a power emphasis coefficient. In the present embodiment, a configuration will be described in which formant enhancement by a linear prediction analysis method is used as a spectrum enhancement technique.

In FIG. 1, 1 is an input voice analysis means for calculating a voice feature parameter from an input voice, 2 is a voice feature parameter variation amount analysis means for analyzing the degree of change of the voice feature parameter, and 3 is the variation. Formant emphasis strength control means for receiving the amount of formant emphasis strength control, 4 is a formant emphasis coefficient map used by the formant emphasis strength control means for determining the emphasis strength, and 5 is formant emphasis in response to the formant emphasis strength control means. Formant emphasis coefficient determining means for determining coefficients, 6 formant emphasis means for performing formant emphasis of input speech, and 7 is a coefficient buffer for storing the time history of the emphasis coefficients determined by the formant emphasis coefficient determining means. Further, 101 is an input voice, 102 is a voice feature parameter, 103 is voice feature parameter variation information, 104 is a formant emphasis temporary coefficient, 105 is a formant emphasis coefficient, 106 is an output voice, 112 is a linear prediction coefficient, and 113 is a previous frame. Is the formant emphasis coefficient of.

Next, the operation will be described. The input voice analysis unit 1 divides the input voice 101 into short time periods, analyzes the input voice for each analysis period, and calculates, for example, an autocorrelation coefficient as the voice feature parameter 102. further,
The linear prediction coefficient 112 is calculated using an autocorrelation method or the like.

The voice feature parameter variation amount analysis means 2 is
The speech feature parameter 102 of the corresponding frame and the past frame is input, and the variation amount of the speech feature parameter in the vicinity of the frame is analyzed. For example, the absolute value of the autocorrelation coefficient and the self-phase with the preceding one frame are analyzed. The change amount of the number of relationships is calculated as the voice feature parameter change amount information 103. It is also possible to perform prefetching to obtain future voice characteristic parameters and use the change in the voice characteristic parameters from the future to the present as the voice characteristic parameter variation amount information 103.

As the voice feature parameter variation amount information 103, in addition to the absolute value of the autocorrelation coefficient and the change amount with respect to the previous frame, for example, the correlation value, the difference value, the differential between the voice feature parameters in the vicinity of the frame concerned. It is also possible to use the value, the absolute value, or the analysis result of the distribution of the phoneme characteristic parameter group by a known statistical analysis method.

The formant emphasizing strength control means 3 comprises a formant emphasizing coefficient map 4 showing the relationship between the form feature emphasizing coefficient and the voice feature parameter variation amount used for determining the formant emphasizing strength as shown in FIG. On the formant enhancement coefficient map 4, the X-axis is associated with the absolute value of the 0th-order component of the autocorrelation coefficient, and the Y-axis is similarly associated with the difference amount of the 0th-order component of the autocorrelation coefficient from the previous frame. XY is obtained by the difference amount between the absolute value of the 0th-order component of the autocorrelation coefficient and the previous frame of the 0th-order component of the autocorrelation coefficient, which is input as the speech feature parameter variation amount information 103 for each short time period.
Since the value of the formant emphasis strength on the Z axis on the plane is uniquely determined, it is output as the formant emphasis temporary coefficient 104.

The formant emphasis coefficient map 4 is provided in a storage device such as a ROM or a RAM. In this embodiment, as a method of creating the method, a method of sequentially changing the formant emphasis coefficient map by a preliminary experiment will be described.
First, several thresholds (Th1, Th2, ..., Thn) on the X-axis that represent the zero-order coefficient of the autocorrelation coefficient and several thresholds (Td1, Td2, ..., Tdn) on the Y-axis that represent the difference amount.
Set. Next, as the emphasis coefficient at the intersection of the meshes on the XY plane formed by the threshold values of the X axis and the Y axis,
At first, an appropriate value is set so as to obtain the optimum formant emphasis coefficient map in terms of hearing, and a temporary speech enhancement device is configured.After that, a listening experiment is performed using the speech enhancement device, and at these intersections, The adjustment of the formant emphasis coefficient is repeated. When it is necessary to subdivide the mesh of thresholds on the X and Y axes of the formant enhancement coefficient map plane in order to perform finer enhancement, a new threshold is set between the thresholds set first, The same adjustment as above is applied to the threshold value. After repeating the above adjustment, the value of the formant emphasis coefficient between the intersections is linearly interpolated with the values of the adjacent intersections. However, in order to prevent oscillation in the formant emphasis means 6, the upper limit value of the formant emphasis coefficient on the formant emphasis coefficient map 4 is set to 0.95, and the lower limit value is set to 0.

The interpolation between the intersections of the formant emphasis coefficient map 4 may be performed by curve interpolation using an interpolation method such as a three-point curve approximation method. Also in this case, in order to prevent oscillation in the formant emphasis means 6, the upper limit value of the formant emphasis coefficient on the map after interpolation is 0.95 and the lower limit value is 0.
Restricted to.

The coefficient buffer 7 is composed of a temporary storage device such as a RAM, has a space for storing the formant emphasis coefficients of the past 5 frames determined by the formant emphasis coefficient determining means 5, and formant emphasis is performed for each short time section. Each time the coefficient 105 is determined, its content is fetched and the value is updated, and the formant emphasis coefficient for the last 5 frames from the current frame is always held.

The formant emphasis coefficient determining means 5 prevents the generation of abnormal noise caused by the discontinuity of the emphasis coefficients with respect to the formant emphasis temporary coefficient 104 determined by the formant emphasis intensity control means 3 due to the discontinuity of the emphasis coefficients with the preceding and following frames. Therefore, the formant emphasis coefficient 113 of the previous frame stored in the coefficient buffer 7 is used, and the five frames are input to a smoothing filter such as a moving average filter to be smoothed. It is output as 105.

The formant emphasizing means 6 emphasizes the formant and the valley portion of the formant (antiformant) so that the transfer function of the filter as represented by the transfer function Hs (z) of the equation (1) is a pole. And a zero-zero model (also called ARMA type) filter. The input voice 101 is input to the filter configured by the equation (1), formant enhancement is performed, and the output voice 106 is output.

[0036]

[Equation 1]

[0037]

[Equation 2]

Α in the equation (1) is based on the formant emphasis coefficient output from the formant emphasis coefficient determining means 5.
The linear prediction coefficient 112 is smoothed according to equation (2). P in the equation (2) is a formant emphasis coefficient, α _o
Is a linear prediction coefficient before smoothing. η further returns the α to the autocorrelation coefficient once, and then weights the autocorrelation coefficient in order to perform coefficient stabilization processing, that is, smoothing the autocorrelation coefficient using a so-called lag window, Further, it is returned to the linear prediction coefficient. For a detailed description of equations (1) and (2), see Ira A Gerson et al., “Vecto.
r Sum Excited Linear Pred
ction (VSELP) Speech Codin
g at 8KBPS ", IEEE, ICASSP,
Since it has been described in (APR, 1990), it will be omitted.
For a detailed description of the lag window, see Japanese Patent Laid-Open No. 52-6.
Since it is described in No. 4808, it will be omitted.

The formant emphasis temporary coefficient 104 prepares a plurality of formant emphasis coefficient maps 4 according to the kind and number of the voice feature parameter change amount information, determines the formant emphasis temporary coefficients in each map, and averages them. It is also possible to determine a representative formant emphasis temporary coefficient.

According to the above-described embodiment, the formant enhancement is performed in response to the variation amount information of the voice characteristic parameter such as the autocorrelation coefficient, so that the rise of the consonant to the vowel causes a rapid change in the spectrum of the voice. There is an effect that an important part can be selectively emphasized with respect to the intelligibility of the voice in the formant of, or the transition part of the formant in which the vowel changes to another vowel. Further, in a transition portion from a vowel to a consonant, etc., by performing emphasis control so as to attenuate the power at the end of the vowel portion earlier than usual, there is an effect of preventing backward consonant masking by the preceding vowel. Further, by preparing a plurality of formant emphasis coefficient maps and determining the formant emphasis coefficient, there is an effect that the formant emphasis processing corresponding to the variation amount of the voice feature parameter can be further performed.

Example 2. A second embodiment of the present invention will be described with reference to FIGS. 3 and 4. In FIG.
Reference numeral 1 is an input voice analysis unit, 2 is a voice feature parameter variation amount analysis unit, 8 is a pre-emphasis coefficient control unit that receives a voice feature parameter variation amount output from the voice feature parameter variation amount unit, and performs pre-emphasis coefficient control,
Reference numeral 9 is a pre-emphasis coefficient determining means for determining the pre-emphasis coefficient in response to the pre-emphasis coefficient controlling means 8, 10 is a pre-emphasis coefficient map used by the pre-emphasis coefficient determining means 9 for determining the emphasis strength, and 11 is a pre-emphasis coefficient map. It is a pre-emphasis means for pre-emphasising the input voice. 102 is a voice feature parameter,
Reference numeral 103 is voice characteristic parameter variation information, 107 is a pre-emphasis temporary coefficient, 108 is a pre-emphasis coefficient,
Reference numeral 114 is a pre-emphasis coefficient of the previous frame. Other input voice 101, coefficient buffer 7, output voice 10
Since 6 is the same as that described in the first embodiment, the description thereof is omitted.

Next, the operation will be described. The input voice analysis unit 1 divides the input voice 101 into short time periods, analyzes the input voice for each analysis period, and calculates, for example, an autocorrelation coefficient as the voice feature parameter 102.

The voice feature parameter variation amount analysis means 2 is
The voice feature parameter 102 of the frame and the past frame is input, and the variation amount of the voice feature parameter in the vicinity of the frame is analyzed. For example, the absolute value of the autocorrelation coefficient and the autocorrelation with the previous one frame are analyzed. The variation amount of the number is calculated as the voice feature parameter variation amount information 103. Further, it is also possible to perform prefetching to obtain a future voice feature parameter and use the change in the voice feature parameter from the future to the present as the voice feature parameter variation amount information 103.

The pre-emphasis coefficient control means 8 is shown in FIG.
The pre-emphasis coefficient map 10 showing the relationship between the pre-emphasis coefficient and the variation amount of the speech feature parameter used for determining the pre-emphasis strength as shown in FIG. On the pre-emphasis coefficient map 10, the absolute value of the 0th order component of the autocorrelation coefficient is associated with the X axis, and the difference amount of the 0th order component of the autocorrelation coefficient from the previous frame is associated with the Y axis. Therefore, XY is obtained by the difference amount between the absolute value of the 0th-order component of the autocorrelation coefficient and the previous frame of the 0th-order component of the autocorrelation coefficient, which is input as the voice feature parameter variation amount information 103 for each short time period. Since the value of the pre-emphasis coefficient on the Z axis on the plane is uniquely determined, it is output as the pre-emphasis temporary coefficient 107.

The pre-emphasis coefficient map 10 is RO
It is provided by a storage device such as M or RAM. In the present embodiment as well, similar to the method of creating the formant emphasis coefficient map shown in the first embodiment, the pre-emphasis coefficient map can be obtained by a method of successively changing the pre-emphasis coefficient map. However, in order to prevent abnormal noise from being generated by the pre-emphasis means 11, the pre-emphasis coefficient map 1
The upper limit of the pre-emphasis coefficient on 0 is 0.7,
The lower limit value is 0.

When the pre-emphasis coefficient map 10 is created by the method shown in the first embodiment, the interpolation between the intersection points on the map may be performed by using an interpolation method such as a three-point curve approximation method. Also in this case, the upper limit of the pre-emphasis coefficient after interpolation is 0.7 and the lower limit thereof is limited to 0 in order to prevent abnormal noise from being generated by the pre-emphasis means 11.

The coefficient buffer 7 is composed of a temporary storage device such as a RAM, has a space for storing the pre-emphasis coefficients for the past 5 frames determined by the pre-emphasis coefficient determining means 9, and pre-emphasis for each short time interval. Each time the coefficient 108 is determined, its value is fetched and the content is updated, and the pre-emphasis coefficient for the last 5 frames from the current frame is always held.

The pre-emphasis coefficient determining means 9 prevents the pre-emphasis temporary coefficient 107 determined by the pre-emphasis coefficient control means 8 from generating an abnormal noise caused by the discontinuity of the emphasis coefficient with the preceding and following frames. For this purpose, the pre-emphasis coefficient 114 of the previous frame stored in the coefficient buffer 7 is used, and the five frames are input to a smoothing filter such as a moving average filter to be smoothed. Output as 108.

The pre-emphasis means 11 is a filter composed of the transfer function H _p (z) expressed by the equation (3). In the pre-emphasis means 11, the pre-emphasis coefficient 108 represented by p and the input voice 101 are input, and pre-emphasis processing of the input voice 101 is performed.
The output voice 106 is output.

[0050]

(Equation 3)

For the pre-emphasis temporary coefficient 107, a plurality of pre-emphasis coefficient maps 10 are prepared according to the type and number of the voice feature parameter change amount information, and the pre-emphasis temporary coefficient is determined in each map. It is also possible to determine the representative pre-emphasis temporary coefficient by averaging.

According to the above embodiment, by performing the pre-emphasis processing in response to the variation amount information of the voice feature parameter such as the autocorrelation coefficient, the consonant to the vowel sound in which the voice feature parameter changes abruptly. There is an effect that the consonant feature whose energy is small and whose power is often concentrated in a high frequency region can be emphasized in the transitional portion of the rising edge, that is, in a portion important for the intelligibility of the voice. Further, by preparing a plurality of pre-emphasis coefficient maps and determining the pre-emphasis coefficient, it is possible to perform the pre-emphasis processing having a higher trackability with respect to the change of the voice characteristic parameter.

Example 3. A third embodiment of the present invention will be described based on FIG. In this embodiment, the spectral emphasis strength control means 3, the formant emphasis coefficient determination means 5, the formant emphasis coefficient means 6, and the pre-emphasis coefficient control means 8 and pre-emphasis coefficient determination described in the second embodiment are described. Both the means 9 and the pre-emphasis means 11 are arranged in cascade.
The respective elements in the figure have been already described in the first and second embodiments and will not be described.

With such a structure, in addition to the effects of the first and second embodiments, the synergistic effect of formant enhancement and pre-emphasis has the effect of further enhancing the high-frequency spectrum peak. Further, in the present embodiment, even if the formant emphasis and the pre-emphasis are switched before and after, the effect of the present invention does not change.

Embodiment 4 FIG. A fourth embodiment of the present invention will be described based on FIG. In FIG. 6, reference numeral 12 is an emphasis incompatible voice detecting means, 13 is an emphasis coefficient changing means, 10
Reference numeral 9 is an emphasis suppression / prohibition signal, and 110 is a changed formant emphasis coefficient.

In the present embodiment, description will be given of a case where the voice emphasizing device described in the first embodiment is applied to the voice emphasizing device 200. Other input speech analysis means 1, speech feature parameter variation analysis means 2, formant emphasis strength control means 3, formant emphasis coefficient map 4, formant emphasis coefficient determination means 5, formant emphasis means 6, coefficient buffer 7, input speech 101, Voice feature parameter 102, voice feature parameter variation amount information 103, formant emphasis temporary coefficient 104,
Since the formant emphasis coefficient 105 and the output voice 106 are the same as those in the above-mentioned embodiment, their explanations are omitted.

Next, the operation will be described. The emphasis non-conforming speech detection means 12 analyzes the speech feature parameter 102 output by the input speech analysis means 1, and when power is concentrated in the low frequency band or when the formant bandwidth is narrower than a certain threshold, the linear prediction is quadratic or lower. When it is discontinued, it is predicted that the output voice will be distorted by emphasizing due to the fact that the noise is large in the voice section (for example, when the SN ratio of the input voice and the noise is less than 10 dB). If so, the emphasis suppression / prohibition control signal 109 is output.

The emphasis coefficient changing means 13 inputs the emphasis suppression / inhibition control signal 109, and when suppressing the formant emphasis, the formant emphasis coefficient 105 output from the formant emphasis coefficient determining means 5 is set as a correction coefficient of 1 The formant emphasis coefficient 110 is output after performing a change process such that the emphasis is weakened by multiplying by a smaller number (here, 0.8), and the formant emphasis coefficient is set to 0 when the formant emphasis is prohibited. Thereafter, the formant emphasizing means 6 performs formant emphasis on the input voice 101 and outputs the output voice 106.
Is output.

By suppressing or prohibiting the formant emphasis, it is possible to prevent the emphasis of the voice that may generate an abnormal sound due to the formant emphasis, suppress the emphasis of noise unrelated to the voice, or to suppress the distortion feeling by the excessive emphasis. Has the effect of preventing

Example 5. A fifth embodiment of the present invention will be described based on FIG. In the present embodiment, as shown in the block diagram of FIG. 7, not only the voice feature parameter 102 but also other parameters, for example, output voice is fed back to the emphasis nonconforming voice detection means 12, and this is analyzed. It is designed to detect excessive emphasis.

By performing feedback control of voice enhancement by using the output voice, there is an advantage that the enhancement control can be performed more precisely.

Example 6. A sixth embodiment of the present invention will be described based on FIG. In the present embodiment, as shown in the block diagram of FIG. 8, an external control signal 111 such as an emphasis intensity changeover switch signal is supplied to the emphasis coefficient changing unit 13 from the outside of the voice emphasis device 201 described in the fourth embodiment. Is input so that the emphasis strength is controlled from the outside.

A new part in FIG. 8 is the external control signal 111. The other elements are the same as the constituent elements described in the fourth embodiment, and the description thereof will be omitted.

By performing the enhancement control using the external control signal, for example, when the voice enhancement device of the present invention is applied to a telephone or the like, the voice enhancement control is performed so that the listener can easily hear the voice enhancement control. Has the advantage that it can be done arbitrarily. Further, there is an advantage that the voice enhancement can be suppressed / prohibited when the listener determines that the voice enhancement is not preferable. On the contrary, it is also possible to perform control such that the suppression / prohibition of voice enhancement is released.

Example 7. In any one of Embodiments 1 to 4, for example, the power of the output sound is normalized so as to be normalized to the power level of the input sound by comparing the power of the current frame of the input sound 101 and the power of the current frame of the output sound 106. You may make it correct.

By performing power correction, for example, it is used in a system in which input voice and output voice are switched and compared with each other, or in a voice storage device capable of emphasizing only a part of a voice message. The power level of the output voice can be matched, and an effect of reducing discomfort at the time of switching can be obtained. Furthermore, there is an effect that an increase in amplitude due to emphasis can be suppressed.

Example 8. In the first, third, and fourth embodiments described above, a constant value bias is applied to the detection result of the voice feature parameter variation amount information 103 so that the voice feature parameter variation amount information 103 always holds a small amount of value. The formant emphasis may be always performed even if the parameters do not change.

There is an advantage that the formant enhancement can be performed not only in the transient part of the phoneme but also in the stationary part such as a vowel whose time variation of the parameter is stable.

Example 9. In the second and third embodiments, a constant value bias is added to the detection result of the voice feature parameter variation amount information 103 so that the voice feature parameter variation amount information 103 always holds a small amount of the voice feature parameter. Pre-emphasis may be performed even if there is no change in parameters.

There is an advantage that pre-emphasis can be performed not only in the transitional part of the phoneme but also in the stationary part such as a vowel whose time variation of parameters is stable.

Example 10. The coefficient map in the above embodiment may be adapted to adapt to changes in the type of voice or changes in environmental noise. Further, as a learning method, for example, a learning means known to those skilled in the art such as an LBG algorithm may be used, or a method of selecting an optimum one from one or more maps may be used. Furthermore, to make it easier for the listener to listen to, for example, when a small amount of emphasis is always desired even when there is no change in the voice feature parameter, an arbitrary change such as adding a small constant to the entire emphasis coefficient of the coefficient map is possible. You may add it. Further, these coefficient maps may be changed in real time independently of the operation of the voice emphasizing device.

By adaptively changing the coefficient map,
It has the advantage that it can follow changes in the environment, changes in the speaker, etc., and can perform optimal voice enhancement for the voice quality of the input voice. Further, there is an advantage that voice enhancement that is easy for the listener to hear can be performed.

[0073]

Since the present invention is configured as described above, it has the following effects.
Since the speech emphasizing device of the present invention derives the spectrum emphasizing coefficient based on the variation amount information of the speech feature parameter, and the spectrum is emphasized on the input speech based on this value, the speech emphasizing unit detects speech such as a transient part of a phoneme. It is possible to maintain the naturalness as a voice and maintain the clarity even in a portion where the spectrum of changes rapidly.

Further, since the speech emphasizing device of the present invention derives the pre-emphasis coefficient based on the variation amount information of the speech characteristic parameter and pre-emphasizes the input speech based on this value, the speech spectrum is sharp. It is possible to maintain the naturalness as a voice and maintain the intelligibility even for a consonant part that changes to a small amount and has a small energy and is often concentrated in a high range.

Further, the speech emphasizing apparatus of the present invention derives the spectral emphasis coefficient and the pre-emphasis coefficient based on the variation amount information of the speech characteristic parameter, and adaptively emphasizes the speech based on both of these values. So the first
And in addition to the effect of the voice emphasizing device of the second invention,
Due to the synergistic effect of the spectral enhancement and the pre-emphasis processing, the spectral peak in the high frequency part of the voice is further enhanced and the clarity can be improved.

Further, the voice emphasizing device of the present invention detects a voice that may generate an abnormal sound by the emphasizing process,
Since the emphasis is suppressed or prohibited, it is possible to suppress the emphasis of noise unrelated to the voice and prevent the sense of distortion due to excessive emphasis.

Further, in the voice emphasizing device of the present invention, since the output voice information is fed back in addition to the input voice to perform the emphasis control, it is possible to obtain a high quality voice output with less discomfort.

Further, since the voice enhancement is performed by using the external control signal, it is possible to perform a flexible voice enhancement process.

Further, since the temporary emphasis coefficient is once obtained from the spectrum emphasis coefficient map, the final coefficient value is further determined from the past history information, and the voice emphasis processing is performed based on this value, so that there is no discomfort. , An output more faithful to the input voice can be obtained.

Further, since the temporary pre-emphasis coefficient is once obtained by the pre-emphasis coefficient map, the final coefficient value is determined from the past history information, and the pre-emphasis processing is performed based on this value. It is possible to obtain an audio output with excellent clarity even for a consonant portion whose energy is low and whose power is often concentrated in a high range.

Furthermore, since a constant bias value is always added to the voice feature parameter variation amount information so that the spectrum of the input voice is always enhanced regardless of whether or not the feature parameter of the input voice varies, not only the transient portion but also the transient portion. Spectral enhancement can be performed even in a stationary part where the voice feature parameter is stable.

Furthermore, since a constant bias value is always added to the voice feature parameter variation amount information so that the pre-emphasis of the input voice is always performed regardless of the variation of the feature parameter of the input voice, not only the transient part but also Pre-emphasis can be performed even in the stationary part where the speech feature parameters are stable.

In addition, since a plurality of spectrum emphasis coefficient maps or pre-emphasis coefficient maps are provided, it is possible to carry out voice emphasis suitable for the voice quality of the input voice that follows changes in the speaker and changes in the environment. .

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention.

FIG. 2 is a diagram showing an example of a formant emphasis coefficient map used in the first embodiment of the present invention.

FIG. 3 is a block diagram showing a second embodiment of the present invention.

FIG. 4 is a diagram showing an example of a pre-emphasis coefficient map used in Embodiment 2 of the present invention.

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

FIG. 6 is a block diagram showing a fourth embodiment of the present invention.

FIG. 7 is a block diagram showing Embodiment 5 of the present invention.

FIG. 8 is a block diagram showing Embodiment 6 of the present invention.

[Explanation of symbols]

1 input speech analysis means, 2 speech feature parameter variation amount analysis means, 3 formant emphasis strength control means, 4 formant emphasis coefficient map, 5 formant emphasis coefficient determination means, 6 formant emphasis means, 7 coefficient buffer, 8
Pre-emphasis coefficient control means, 9 pre-emphasis coefficient determination means, 10 pre-emphasis coefficient map, 1
1 pre-emphasis means, 12 emphasis incompatibility voice detection means, 13 emphasis coefficient changing means, 101 input speech, 1
02 voice feature parameter, 103 voice feature parameter variation amount information, 104 formant emphasis temporary coefficient, 105
Formant emphasis coefficient, 106 output speech, 107 pre-emphasis temporary coefficient, 108 pre-emphasis coefficient,
109 emphasis suppression / inhibition signal, 110 modified formant emphasis coefficient, 111 external control signal, 112 linear prediction coefficient, 113 previous frame formant emphasis coefficient, 1
14 Pre-emphasis coefficient of previous frame, 200 Speech enhancement device described in Example 1, 201 Speech enhancement device described in Example 4.

Claims (11)

[Claims] 1. A speech feature parameter variation amount analyzing means for analyzing a degree of temporal variation of a speech feature parameter of an input speech divided into short time intervals, and a speech feature parameter variation output by the speech feature parameter variation amount analyzing means. Spectral emphasis strength control means for determining a spectral emphasis strength and outputting a spectrum emphasis coefficient in response to the quantity information, and a spectrum emphasis means for performing a spectrum emphasis of the input voice according to the spectrum emphasis coefficient output by the spectrum emphasis strength control means. And a voice emphasizing device. 2. A voice feature parameter variation amount analyzing means for analyzing a degree of time variation of a voice feature parameter of an input voice divided into short time intervals, and a voice feature parameter variation output by the voice feature parameter variation amount analyzing means. Pre-emphasis coefficient control means for determining a pre-emphasis coefficient in response to the quantity information and outputting the pre-emphasis coefficient; and pre-emphasis means for pre-emphasis of the input voice according to the pre-emphasis coefficient output by the pre-emphasis coefficient control means. And a voice emphasizing device. 3. A voice feature parameter variation amount analyzing means for analyzing a degree of time variation of a voice feature parameter of an input voice divided into short time periods, and a first feature output from the voice feature parameter variation amount analyzing means.
Spectral emphasis strength control means for determining the spectral emphasis strength and outputting a spectral emphasis coefficient in response to the speech feature parameter variation amount information of the input speech, and the spectral emphasis of the input speech according to the spectral emphasis coefficient output by the spectral emphasis strength control means. And a second feature output by the voice feature parameter variation analysis unit.
Pre-emphasis coefficient control means for determining the pre-emphasis coefficient and outputting the pre-emphasis coefficient in response to the speech feature parameter variation amount information, and the pre-emphasis of the input speech according to the pre-emphasis coefficient output by the pre-emphasis coefficient control means. And a pre-emphasis means for performing the above. 4. An emphasis nonconforming voice detection means for inputting a voice feature parameter of an input voice divided into short time periods and outputting an enhancement suppression signal when distortion is expected in the output voice, and based on the enhancement suppression signal. 2. An enhancement coefficient changing unit for correcting the spectrum enhancement coefficient according to claim 1, wherein the spectrum enhancement unit is configured to perform the spectrum enhancement of the input voice by the corrected spectrum enhancement coefficient. Speech enhancement device. 5. The emphasis incompatibility voice detection means controls the emphasis suppression signal from the voice feature parameter information of the output voice in addition to the voice feature parameter information of the input voice. Voice enhancement device. 6. The speech enhancement apparatus according to claim 4, wherein the enhancement coefficient changing means determines the spectral enhancement coefficient by an external control signal in addition to the enhancement suppression signal. 7. A spectrum emphasis coefficient map that defines a relationship between speech feature parameter variation amount information and a spectrum emphasis coefficient, and a spectrum emphasis temporary coefficient determination unit that calculates a spectrum emphasis coefficient from the spectrum emphasis temporary coefficient. The intensity control means outputs a value derived from the spectrum emphasis coefficient map based on the speech feature parameter variation amount information to the spectrum emphasis coefficient determination means as a spectrum emphasis temporary coefficient, and the spectrum emphasis coefficient determination means determines the spectrum emphasis coefficient. The speech enhancement apparatus according to claim 1, wherein a spectrum enhancement coefficient determined by smoothing processing of preceding and following speech frames is calculated from the temporary coefficient and output to the spectrum enhancing means. 8. A pre-emphasis coefficient map that defines a relationship between voice feature parameter variation information and a pre-emphasis coefficient, and a pre-emphasis coefficient determination unit that calculates a pre-emphasis coefficient from a pre-emphasis temporary coefficient. The control means outputs a value derived from the pre-emphasis coefficient map based on the voice feature parameter variation information to the pre-emphasis coefficient determining means as a pre-emphasis temporary coefficient, and the pre-emphasis coefficient determining means determines the pre-emphasis temporary coefficient. 3. The speech emphasizing device according to claim 2, wherein a pre-emphasis coefficient determined by smoothing the speech frames before and after the coefficient is calculated from the coefficient and output to the pre-emphasis means. 9. The speech feature parameter variation amount information is constantly added with a constant bias value so that the spectrum of the input speech is always enhanced regardless of the variation of the characteristic parameter of the input speech. The speech enhancement device according to any one of paragraphs 1 or 3 to 7. 10. The pre-emphasis of the input voice is always performed regardless of the change of the feature parameter of the input voice by always adding a constant bias value to the voice feature parameter variation amount information. The speech enhancement device according to item 2, item 3, or item 8. 11. The speech enhancement apparatus according to claim 7, wherein a plurality of the spectral enhancement coefficient maps and the pre-emphasis coefficient maps are provided.