KR101682796B1 - Method for listening intelligibility using syllable-type-based phoneme weighting techniques in noisy environments, and recording medium thereof - Google Patents

Abstract

A method for enhancing speech intelligibility using a syllable-form based phoneme weighting technique in a noisy environment, the method comprising the steps of: detecting a syllable from a speech signal; classifying the syllable form by analyzing the detected syllable; Adjusting the power of each consonant according to the syllable type in the environment, and outputting the power of the voice signal having the power of the consonant and the power of the voice signal not having the power equalized through power normalization. According to the present invention, it is possible to improve intelligibility in a noisy environment by weighting phonemes based on the syllable form.

Description

BACKGROUND OF THE INVENTION Field of the Invention The present invention relates to a method for enhancing intelligibility of a speech using a syllable form based phoneme weighting technique in a noisy environment,

The present invention relates to phoneme weighting techniques based on syllable form in a noisy environment.

Speech or announcement speech delivered to a listener at a lecture hall or a train station may greatly reduce the intelligibility of speech due to ambient noise. This situation also occurs in the case of watching a TV or making a mobile phone voice call. In a listening environment where ambient noise is present, the voice signal is masked by ambient noise, making it difficult for the listener to perceive some voice signals, especially for the hearing impaired and the elderly. Since the masked speech signal is difficult for the listener to recognize some speech signals, a technique for enhancing the clarity of the speech by enhancing the speech signal at the speech output stage is needed.

Speech enhancement, or noise suppression, has been studied to improve the intelligibility of speech in a noisy environment. However, it has been reported that noise can not be directly controlled, that is, when a listener is not wearing a noise attenuator It is not applicable in the situation.

In general, the method of increasing the signal-to-noise ratio (SNR) or the speaker volume in order to improve the clarity of the voice has a problem that all components of the output voice unrelated to the clarity are constantly increased, May be degraded or distorted, which may cause a fatigue and a sense of rejection of the listener.

Another method is to enhance the power of consonants. Consonants convey more important information than vowels in terms of clarity of speech, and consonants are less powerful than vowels, making it difficult to listen in a noisy environment.

The consonant-vowel intensity ratio (CVR) technique can be used to enhance the power of consonants. The CVR increases the sound pressure level (SPL) of the consonant as a power ratio of the consonant to the vowel, and is known to be advantageous for consonant listening as the CVR level increases. However, the enhancement of simple consonants through the CVR technique may not significantly improve speech intelligibility. Although the opinion that the listening ability of consonants improves with increasing CVR level is mostly dominant, opinions on the effect of word recognition score according to CVR technique are still not consistent.

In a related study to confirm the effect of word recognition according to the CVR technique in Korea, consonant-vowel (CV) syllable which combines consonants difficult to hear in Korean with several vowels was examined by CVR As a result, the awareness difference does not improve significantly with increasing CVR levels. This means that only some consonants of consonants are effective in improving awareness, and applying the CVR technique which over-reinforces consonant power does not necessarily help to improve awareness of all listeners. It is difficult to observe the effect of the CVR technique which enhances consonant only in many aspects. Therefore, it is necessary to consider how the awareness changes by using human auditory perception characteristics fundamentally.

In Korean, the vowels are mainly distributed in the low frequency region, are highly related to the voice energy, and the consonants are mainly distributed in the high frequency band and have a great influence on the intelligibility. However, the frequencies of the consonants are different in frequency range depending on the trailing short vowels. For example, the average center frequency of the consonant / ㅅ / is about 6,200 Hz, and / ㅆ / is about 6,600 Hz, but the short center / 오 / When / / is / is followed, the frequency range of / ㅅ / is 4,044 to 6,461 Hz and / ㅆ / is 4,357 to 6,767 Hz.

The vowel is a phoneme that should be considered in the clarity of the voice, since it means that the awareness of the listener can be changed according to the vowel even if it is a consonant corresponding to the high frequency region which is difficult to be perceived in such an acoustical cue. Therefore, consonant-vowel-consonant (CVC), CV, vowel-consonant (VC) and vowel (V) syllables composed of two or more phonemes in Korean syllable form Since the frequency range of each type is different, the awareness may vary.

Korean Patent Publication No. 10-2007-0112848

SUMMARY OF THE INVENTION It is an object of the present invention to provide a consonant-vowel intensity ratio (CVR) technique that uniformly enhances the power of all consonants, Consonant-vowel-consonant (CVC), consonant-vowel (CV), vowel-consonant (VC), and vowel The present invention aims to provide a syllable form based phoneme weighting method for a clear listening in a noisy environment.

The objects of the present invention are not limited to the above-mentioned objects, and other objects not mentioned can be clearly understood by those skilled in the art from the following description.

According to an aspect of the present invention, there is provided a method for enhancing the intelligibility of a speech using a syllable type based phoneme weighting technique in a noisy environment, comprising: detecting a syllable from a speech signal; classifying the syllable type by analyzing the detected syllable Adjusting the power of the consonants according to the syllable type in the noise environment estimated from the ambient noise, and outputting the power of the voice signal having the power of the consonant enhanced and the power of the voice signal not having the power equalized through power normalization .

The syllable forms include consonant-vowel-consonant (CVC), consonant-vowel (CV), vowel-consonant (VC) and vowel .

In the embodiment of the present invention, the rating expectation values of the syllable types are obtained based on the evaluation result of the monosyllable speech recognition, and the power of the consonants can be adjusted by using the recognition degree expectation value for each syllable type.

(수학식 3)으로 나타낼 수 있다. X is a random variable indicating the grade of each syllable form V, VC, CVC and CV, and E [X] is the SPL (sound pressure level) adjustment value (dB) for phoneme weighting Quot;

(3). &Quot; (3) "

(수학식 4)로 나타낼 수 있다. Each syllable, each type differently applied weight range is determined by the dynamic range (dynamic range) between the △ _CW y y _CV and _V,

(Equation 4).

(수학식 5)와 같이 나타낼 수 있다. The constant a,

(5). &Quot; (5) "

(수학식 6)으로 나타낼 수 있으며, 상수 b는 상기 △P_c와 평균적으로 같아지도록,

(수학식 7)로 나타낼 수 있다.

Can be represented by (Equation 6), b is a constant to be equal and the average △ P _c,

(7). &Quot; (7) "

In one embodiment of the present invention, the [Delta] _CW may be 12 [dB].

According to the present invention, it is possible to improve intelligibility in a noisy environment by weighting phonemes based on the syllable form. That is, according to the present invention, the performance is superior to that of the conventional CVR technique and the speech intelligibility is improved.

FIG. 1 is a block diagram showing a CVR technique for improving speech intelligibility.
FIG. 2 is a block diagram illustrating a method for enhancing the intelligibility of speech using a syllable type based phoneme weighting technique according to an embodiment of the present invention. Referring to FIG.
FIG. 3 is a chart showing a grade expectation value according to an embodiment of the present invention.
FIG. 4 is a graph showing a sound pressure level increase value of a phoneme according to an embodiment of the present invention.

While the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It is to be understood, however, that the invention is not to be limited to the specific embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.

The terminology used in this application is used only to describe a specific embodiment and is not intended to limit the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In the present application, the terms "comprises" or "having" and the like are used to specify that there is a feature, a number, a step, an operation, an element, a component or a combination thereof described in the specification, But do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, or combinations thereof.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in commonly used dictionaries are to be interpreted as having a meaning consistent with the contextual meaning of the related art and are to be interpreted in an ideal or overly formal sense unless expressly defined in the present application Do not.

In the following description of the present invention with reference to the accompanying drawings, the same components are denoted by the same reference numerals regardless of the reference numerals, and redundant explanations thereof will be omitted. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description, well-known functions or constructions are not described in detail since they would obscure the invention in unnecessary detail.

In the present invention, the subject performing the speech intelligibility enhancement method using the syllable type based phoneme weighting method may be a system for outputting speech, or may be a system for controlling a system or an apparatus for outputting speech, ). That is, the method of enhancing the intelligibility of the speech using the syllable type based phoneme weighting technique of the present invention is composed of an algorithm which is a kind of software, and the software can be executed in a sound output system, a control unit of the sound output apparatus, or a processor.

FIG. 1 is a block diagram showing a CVR technique for improving speech intelligibility.

Referring to FIG. 1, a conventional consonant-vowel intensity ratio (CVR) technique for clear listening in a noisy environment estimates noise 120 from ambient noise, detects consonants at a speech output stage 110, The power of the consonants is increased uniformly in proportion to the magnitude of the noise (130). For example, according to the estimated noise level, the power of the consonant is increased by +3 dB when the noise is small and by +12 dB when the noise is large.

Thereafter, power of the consonant is increased by the power normalization 140, and the power of the voice signal of which the power of the consonant is enhanced is made equal to the power of the voice signal of no processing.

In the CVR technique, a weight is applied to the consonant signal x _c in the power P _c of the consonant signal to calculate the weight so that the power P ' _c of the enhanced consonant signal can be derived.

Speaking of the original α a weight multiplied to the consonant signal, x _c Substituting instead αx _c is expressed as formula (1).

Here, the weight value? Can be obtained by the following equation (2).

Here, ΔP _c is the SPL increase value of the consonant, that is, P ' _c -P _c .

Unlike the conventional CVR technique which uniformly enhances the power of all consonants for clear listening in a noisy environment, the present invention proposes a corrective weighting (CW) technique for enhancing the power of each consonant separately for each syllable type .

FIG. 2 is a block diagram illustrating a method for enhancing the intelligibility of speech using a syllable type based phoneme weighting technique according to an embodiment of the present invention. Referring to FIG.

Referring to FIG. 2, the syllable type based phoneme weighting method of the present invention first detects a syllable from a speech signal (210).

Then, the detected syllable is analyzed to sort syllable types (220).

Then, noise is estimated from ambient noise (230).

Then, the power of the consonants is adjusted differently according to the syllable type in the estimated noise environment (240).

Thereafter, the power normalization unit 250 outputs the power of the consonant with the same power as that of the voice signal with the enhanced power of the consonant.

The present invention is based on consonant-vowel-consonant (CVC), consonant-vowel (CV), vowel-consonant (VC), vowel- , And V), thereby providing a syllable-form-based phoneme-weighting method for clear listening in a noisy environment.

For the syllable form-based phoneme weighting technique of the present invention, 250 test notes were selected from the analysis of the frequency of use of Korean morpheme and vocabulary in consideration of the frequency, and as a result, The recognition rating is classified into 11 grades. As the grade approaches 1, it is a syllable of the highest rank with a high recognition rate. In the upper class, the syllables are uniformly distributed, but in the lower class, CVC and CV are more distributed. Therefore, since CVC and CV are syllable forms with low clarity, it is necessary to further strengthen the power of consonants compared to other syllable forms.

In the present invention, the grade expectation value of each syllable form can be obtained based on the evaluation result of the monosyllabic pronunciation recognition, and the result can be confirmed in the chart of FIG.

FIG. 3 is a chart showing a grade expectation value according to an embodiment of the present invention.

The SPL (Sound Pressure Level) adjustment value (dB) y for the phoneme weighting by the syllable type is determined linearly and can be expressed by the following equation (3).

Where X is a random variable representing the grade of each syllable form V, VC, CVC, and CV, and E [X] is the grade expectation.

And, each syllable each type differently applied weight range is determined by the dynamic range (dynamic range) between the _CW △ y and y _{V CV.}

Here,? _CW is experimentally obtained and is 12 dB.

Then, a constant a to be multiplied by the grade expectation for each syllable type is expressed by Equation (5).

Here, the constant b is obtained as follows so as to be equal to ΔP _{c in} Equation 2 on average.

FIG. 4 is a graph showing a sound pressure level increase value of a phoneme according to an embodiment of the present invention.

Referring to FIG. 4, the results of calculation of the SPL increase value by the syllable type are summarized, and it is understood that the SPL increase values by the VC, CVC, and CV syllable types are different from each other.

Adjusting the SPL increase in syllable form V so that? _CW is 12 dB as in Equation 8 below, then yV is -3.78 dB.

Meanwhile, the method of enhancing the intelligibility of speech using the syllable type based phoneme weighting technique in the noisy environment according to the embodiment of the present invention can be implemented as a computer readable code on a computer readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored.

For example, the computer-readable recording medium includes a ROM, a RAM, a CD-ROM, a magnetic tape, a hard disk, a floppy disk, a removable storage device, a nonvolatile memory, , And optical data storage devices.

In addition, the computer readable recording medium may be distributed and executed in a computer system connected to a computer communication network, and may be stored and executed as a code readable in a distributed manner.

While the present invention has been described with reference to several preferred embodiments, these embodiments are illustrative and not restrictive. It will be understood by those skilled in the art that various changes and modifications may be made therein without departing from the spirit of the invention and the scope of the appended claims.

210 Syllable detection 220 syllabic classification
230 Noise estimation 240 Phoneme power control
250 Power normalization

Claims (9)

A method for enhancing speech intelligibility using a syllable form based phoneme weighting technique in a noisy environment,
Detecting a syllable from a speech signal;
Analyzing the detected syllable and classifying the syllable form;
Adjusting the powers of the consonants differently according to syllable forms in a noise environment estimated from ambient noise; And
Outputting power of a voice signal in which power of a consonant is enhanced through power normalization and power of a voice signal not having the same power,
The syllable form includes a consonant-vowel-consonant (CVC), a consonant-vowel (CV), a vowel-consonant (VC), and a vowel ,
In the step of adjusting the powers of the consonants according to syllable types in the noise environment estimated from the ambient noise,
Based on the evaluation result of the monosyllabic speech recognition, the recognition degree expectation value for each syllable type is obtained,
Wherein the power of the consonants is adjusted differently by using the recognition expectancy of the syllable type according to the syllable type based phoneme weighting technique.
delete delete The method according to claim 1,
X is a random variable indicating the grade of each syllable form V, VC, CVC and CV, and E [X]
The SPL (Sound Pressure Level) adjustment value (dB) y for the phoneme weight by the syllable type of each syllable type,

(Equation 3). ≪ / RTI >< RTI ID = 0.0 >
The method of claim 4,
Each syllable by type different syllable phoneme format specific weight range to be applied is determined by the dynamic range (dynamic range) between the △ _CW y y _CV and _V,

(4). ≪ / RTI > A method for enhancing the intelligibility of speech using a syllable type based phoneme weighting technique.
The method of claim 5,
The constant a,

(Equation 5). ≪ / RTI > In the method of enhancing the intelligibility of speech using the syllable-form-based phoneme weighting technique,
delete The method of claim 5,
And [Delta] _CW is 12 [dB]. A method for improving intelligibility of speech using a syllable form-based phoneme weighting technique.
A computer-readable recording medium having recorded thereon a program capable of causing a computer to execute the method according to claim 1, claim 4, claim 6 or claim 8.

Images