KR20080087541A - System and method for synthesizing english voice using korean alphabet - Google Patents

Abstract

An apparatus and a method for synthesizing English voice by using Korean alphabets are provided to achieve three or more continuous consonant initial sound inputs, thereby achieving input of consonant bundle and synthesizing the English voice closer to a pronunciation of a native speaker. An apparatus(100) for synthesizing English voice by using Korean alphabets comprises the following. A Korean alphabet input unit(110) receives input of Korean alphabet. A voice converting unit(120) converts the Korean alphabet input into corresponding predetermined English voice. An English voice output unit(130) outputs the English voice. The Korean alphabet input unit performs an initial sound input formed with three or more continuous consonants.

Description

System and method for synthesizing English voice using Korean alphabet}

1 is a schematic block diagram of an English speech synthesis apparatus using Hangul according to the present invention.

2 is a diagram illustrating an intraoral reflection point of an unvoiced friction sound.

Figure 3 shows the relative position of the tongue of the silent friction sound.

Figure 4 shows the position of the tongue of 'th' pronunciation.

5 is a view showing a position where the tongue of the 'ㄹ' pronunciation touches the palate.

Figure 6 shows the shape of the tongue of the 'l' pronunciation.

Fig. 7 shows the shape of the tongue of 'r' pronunciation.

r' 발음의 혀의 중간선 위치를 도시한 도면.8 is' er 'and'

r 'pronunciation midline position of tongue.

'의 상대적 발음 위치를 도시한 도면.9a shows ah and ah

A diagram showing the relative pronunciation position of '.

'의 상대적 발음 위치를 도시한 도면.Figure 9b is' o 'and'

A diagram showing the relative pronunciation position of '.

', '

'의 상대적 발음 위치를 도시한 도면.10 is' er 'and'

','

A diagram showing the relative pronunciation position of '.

Fig. 11A shows the pronunciation positions of 'u' and 'u'.

Fig. 11B shows the pronunciation positions of 'u' and 'u'.

12A illustrates the shape of a rigid tongue when pronounced 'i ^y ' and 'u ^w '.

12B is a view showing relative pronunciation positions of Korean 'I' and 'I' and 'i ^y ' in English.

FIG. 13 is a view showing an adhesive site of a tongue when 'tl' is pronounced. FIG.

14 is a view showing the adhesive portion of the tongue when pronounced 'tn'.

15A illustrates the movement of the tongue when pronounced 't'.

Fig. 15B shows the movement of the tongue when pronounced 'tr'.

16 is a consonant table written according to the notation method according to an embodiment of the present invention.

17 is a vowel table marked according to the notation method according to an embodiment of the present invention.

The present invention relates to a speech synthesis apparatus and a method, and more particularly, to a speech synthesis system and method that can generate a pronunciation close to the pronunciation of a foreign language using the Hangul alphabet.

The voice synthesizer refers to a device for synthesizing and uttering a voice signal by electronic means. The voice synthesizer is mainly used in a voice response device and the like, and generally receives text data and converts the voice data into voice data.

With the trend of internationalization, the use of English is becoming more common in everyday life, and accordingly, there is a need to synthesize English voices in Korea.

However, since Hangul is a character optimized for pronunciation of Korean, when a Korean is synthesized by inputting Korean in a conventional manner, a native speaker may not clearly understand the meaning of the synthesized English voice due to awkward pronunciation. .

The present invention has been made to solve the above-described problems, and an object of the present invention is to provide an apparatus and method for synthesizing an English voice closer to a native speaker's pronunciation using Korean characters.

In order to achieve the above object, an English speech synthesizer using Korean according to the present invention includes a Korean input unit, a speech converter, and an English speech output unit.

The Hangul input unit receives a Hangul input, and may perform an initial input formed of three or more consecutive consonants. The voice converter converts the Hangul input into a predetermined English voice, and the English voice output unit outputs the converted English voice.

By being able to input three or more consonant initial consonants, it becomes possible to input a consonant bundle as in English, thereby synthesizing an English voice closer to the native speaker's pronunciation.

In addition, the Hangul input unit may perform input of a dependent vowel distinct from the main vowel. By using Hangul, it is possible to input dependent vowels as in English, thereby synthesizing English voices closer to the pronunciation of native speakers.

In addition, an invention claiming the device invention in the form of a method is disclosed.

Hereinafter, exemplary embodiments of the present invention will be described with reference to the accompanying drawings.

1 is a schematic block diagram of an English speech synthesis apparatus using Korean according to the present invention.

In FIG. 1, the speech synthesis apparatus 100 includes a Hangul input unit 110, a voice converter 120, and an English voice output unit 130.

The Hangul input unit 110 receives a Hangul input and may perform an initial input formed of three or more consecutive consonants. The voice converter 120 converts the Hangul input into a predetermined English voice, and the English voice output unit 130 outputs the converted English voice. In addition, the Hangul input unit 110 may input a dependent vowel distinguished from the main vowel.

In this way, the Korean consonant bundle and the dependent vowel input in the English language can be input using the Korean language, so that the English voice can be synthesized closer to the pronunciation of the native speaker using the Korean language.

There are more than 40 phonemes (consonants and vowels) in English, and only a few of them exactly match the phonemes in Korean. In addition, the phoneme combination method for forming a word (word) also has a big difference from Korean.

Therefore, in order to pronounce English relatively accurately, the pronunciation position and utterance of each phoneme (consonants and vowels), and the utterance according to the combination of the phonemes are very important, and the importance of accent cannot be overemphasized.

Hereinafter, first, the pronunciation position and the utterance method of each phoneme will be described, and then the phonation method according to the combination of the phonemes will be described.

Silent rupture [k, p, t]

In general, the burst sound immediately preceding the first and vowels (accent vowels) of the word is pronounced unvoiced. However, in case of (s + rupture sound), unvoiced rupture is impossible.

It's too hard to pronounce the silent s, which is blown by the wind at high speed, and compress air instantly for a burst. Therefore, weak burst pronunciation is made. Weak burst pronunciation is a pronunciation that is weaker than the strong burst pronunciation and pronounced at the same position as the strong burst pronunciation.

The weak burst pronunciation is divided into a silent weak burst pronunciation and a voiced weak burst pronunciation. If the meteor sound is followed by the weak burst pronunciation, it will be meteorized. In American English, the vocalization burst pronunciation is applied by applying the same principle as in the case of (gang vowel + rupture + weak vowel). American English tends to pronounce bristles with faster air than British.

, h]Friction sound [f, v, θ, ð, s, z, ∫,

, h]

Friction is the sound of blowing wind through narrow gaps of about 1mm. The mouth and tongue are pronounced in a fixed state. In Korean, there are three (ㅅ, ㅆ, ㅎ). In special cases, when you ask to be quiet, you can raise the perception of '∫' while touching your lips. In other words, it is not 'sh ~' and '∫ ~' without vowel pronunciation. Therefore, the friction sound, like in Korean, is generally 'h'.

When our friction sound comes to the base, it turns into a non-ruptured pronunciation. However, English does not have a base, so always pronounces the friction sound. Unvoiced friction sounds by blowing air much faster than voiced friction.

At this time, the fast wind passing through the narrow gap hits and reflects at another point in the mouth, so the sound is amplified and we hear it. Therefore, knowing the reflection point of the unvoiced friction sound helps in correct pronunciation.

2 is a view showing a reflection point in the oral cavity of the silent friction sound, Figure 3 is a view showing the relative position of the tongue.

]TH [θ,

]

Spread your tongue sideways and close the edge of your tongue to the crown of your upper teeth. At this time, the tip of the tongue did not touch the teeth and cut about 1mm.

4 is a view showing the position of the tongue when the 'th' pronunciation.

L [l]

There are two pronunciations of ㄹ in Korean. First, 'rae' in the 'song' is the medium-type r, 'ra' in the song 'r' can be referred to as the typical r.

5 is a view showing a position where the tongue of the 'ㄹ' pronunciation touches the palate.

In the typical d, the tip of the tongue touches the alveolar, while the middle d is in contact with the mouth ceiling of the tongue, but not the upper teeth of the upper teeth.

There are two kinds of English 'l' pronunciations. L before the vowel and l pronounced consonant after the vowel. The l before the vowel is the same as the Korean typical ㄹ.

6 is a view showing the shape of the tongue when pronounced 'l'.

The shape of the tip of the tongue is as shown in Figure 6a, as can be seen in Figure 6b, when looking at the flow of the air of the typical and after the typical air flow comes out to the front edge than the rear.

R [r]

The pronunciation of the consonant 'r' comes only in front of the vowel. The most important thing to pronounce r is to touch the edge of the tongue to the point where the upper teeth and gum meet, except for the tip of the tongue. Never try to roll it up, and the centerline of the tongue is slightly lower than the edges.

7 is a view showing the shape of the tongue when the consonant 'r' is pronounced. The shape of FIG. 7 is similar to that of only the tip of the tongue in the shape of the medium d of FIG. But the middle back of the tongue is lower.

The distance between the tip of the tongue and the palate is about 1 mm. In the upper position, the consonant r is pronounced, moving the tongue tip away from the palate.

r'은 혀끝에서 소리난다. 혀끝이 치경부의 0.5cm 뒤쪽을 가리킨다. 혀끝은 경구개에서 2mm 정도 떨어진다. 너무 가까우면 마찰음이 되기에 주의한다. 입술 은 한국어의 어 정도 벌어진다. 특히 이는 단모음이다. 혀가 중간에 특히 어 있는 부분에 솟아오르면 안된다.'

r 'sounds at the tip of the tongue. The tip of the tongue points 0.5cm behind the cervical region. The tip of the tongue is about 2 mm from the palate. Be careful when you get too close to the sound of friction. Lips open in Korean. In particular it is a short vowel. Your tongue should not rise above the middle.

r' 발음의 혀의 중앙선 위치를 도시한 도면이다. 도시된 바와 같이, 혀의 가장자리는 자음 r과 같이 이빨과 잇몸이 만나는 부분에 닿게 한다.8 is' er 'and'

It is a figure which shows the position of the center line of the tongue of r 'pronunciation. As shown, the edge of the tongue touches the area where the teeth and gum meet, such as consonant r.

] (law, paul, saw, coffee) [

(law, paul, saw, coffee)

'　'Ah'Vs'

'

'의 상대적 발음 위치를 도시한 도면이다.9a shows ah and ah

'Is a diagram showing a relative pronunciation position.

As shown in the figure, the mouth is lower than the 'a' in Korean and the mouth is wider than the 'a'.

'　'Oh Vs'

'

'의 상대적 발음 위치를 도시한 도면이다.Figure 9b is' o 'and'

'Is a diagram showing a relative pronunciation position.

As shown in the figure, 'oh' sounds from the tip of the tongue.

', '

''Er'Vs'

','

'

', '

'의 상대적 발음 위치를 도시한 도면이다.10 is' er 'and'

','

'Is a diagram showing a relative pronunciation position.

'는 혀가 윗니나 잇몸에 닿지 않는다. 그러나 '

' 는 한국어의 '에' 또는 '으'처럼 혀의 가장자리가 윗니 크라운에 닿거나 거의 닿을 듯이 하여 발음한다. '

'는 뒤따라오는 자음으로 움직이기에 가장 좋은 위치의 모음이다. 또한, 가장 힘을 들이지 않고 발음할 수 있는 모음이다.Our 'er' and the 'english'

The tongue does not touch the upper teeth or gums. But '

'Is pronounced as if the edge of the tongue touches or almost touches the upper teeth crown, like' E 'or' U 'in Korean. '

'Is a collection of the best positions to move with the following consonants. It is also a vowel that can be pronounced with the least effort.

U [u] (took, book, pull)

'U' Vs 'u'

11A is a diagram showing pronunciation positions of 'u' and 'u'.

The closest pronunciation of 'u' is 'U' in Korean.

'U' Vs 'u'

11B is a diagram illustrating pronunciation positions of 'right' and 'u'.

Vowels are determined at the narrowest position. Be careful if it is too narrow as it will be a friction sound.

^y i _(this), u ^w _{(coming from} the right)

In Korean, there is no vowel sounding the tongue. But the long vowels of English i ^y (beat, see) and u ^w (pool, you) pronounce the tongue straining.

12A is a diagram illustrating the shape of a rigid tongue when pronounced 'i ^y ' and 'u ^w '.

As shown in FIG. 12A, when the tongue is tense, a minute swelling occurs in the area, and the tongue is closer to the palate. The reaction also makes the lower part of the tongue push down under the chin.

The main vowels are pronounced while the tongue is in a relaxed state and the short lingering sound that occurs when the tongue is slightly far from the palate when it returns to the relaxed state is a dependent vowel.

12B is a diagram illustrating the relative pronunciation positions of Korean 'I' and 'I' and 'i ^y ' in English.

''ㄱ, ㄷ, ㅂ, ㅈ'Vs' g, d, b,

'

The air flow to pronounce unvoiced sounds is much faster than voiced sounds. In Korean, a, c, ㅂ, ㅈ is pronounced if it is in the middle of a word or a group of words. But when these are the first notes of a word or group of words, they are pronounced unvoiced using a fast stream of air. This is why 'Busan' in Korean sounds 'Pusan' to foreigners.

'로 시작하는 단어를 빠른 공기의 흐름으로 발음하면 'k, t, p,

또는 t'로 오해받게 된다. 따라서 느린 공기의 흐름으로 가볍게 시작하면 유성음 자음의 발음을 쉽게 할 수 있다.Voiced 'g, d, b,

A word beginning with "is pronounced as a stream of fast air, where" k, t, p

Or misunderstood t '. Therefore, if you start lightly with a slow air stream, you can easily pronounce the voiced consonants.

N, T, D [n, t, d]

The part of the tongue adhering to the palate for the pronunciation of the three phonemes is shown in FIG. 8. This is the position where the tongue sticks to the palate when the mouth is closed. In general, the pronounced 't' strong burst, weak burst, meteorological burst, and non-rupture all start at the same location.

Meteor burst t + l (St. Louise, settle, seatle)

The position of the tongue starts as shown in FIG. 11, while the tip of the tongue is attached to the alveolar, pushing the rest of the tongue strongly against the palate and pronounces t and l while leaving the rest of the tongue. At this time, the back of the tongue falls from the palate is t.

Fig. 13 is a view showing the adhesion site of the tongue of 'tl'.

Oily burst t + n (Partner, Clinton)

Stick the back of the tongue a little longer than usual to pronounce 't' to prevent the flow of air to the nose to the soft palate and open the palate strongly and pronounce the t and n with the front of the tongue intact. At this time, the sound of opening of the dog is 't'.

Fig. 14 is a view showing the adhesion site of the tongue of 'tn'.

W [w] (swing, queen, quiz, twin, squid)

The phoneme is almost the same as the pronunciation of 'u' in 'up' or 'o' in Korean. If w comes before the vowel, it is used as a consonant. Therefore, the consonants before w should be pronounced independently.

Pronunciation method by phoneme combination

The basic syllables of Korean words (words) are in the form of {(consonant) + vowel + (support)}. There is no need for consonants or support here. In other words, the basic element of syllable is vowel.

One syllable word in English is in the form of {(consonant bundle) + vowel + (consonant bundle)}, and the consonant bundle consists of several consonants. There are short vowels and double vowels in English vowels.

Two-syllable words in English are in the form of {(consonant bunch) + vowel + (consonant bunch) + vowel + (consonant bunch)}, and the pronunciation of the vowel when the word ends with l, m, n in more than one syllable This may be omitted.

In Korean, there are only a few consonants. How to pronounce a bunch of consonants with only a few consonants is very important in English. Consonant consonants are pronounced all consonants in turn, taking advantage of the characteristics of each consonant, as if the entire consonants should be pronounced. To do this, the following tips and laws are required.

Law of Support of Vowels

English does not have consonants as a backing. The pronunciation of vowels with and without vowels in Korean is different. The beginning of pronunciation is the same, but the vowels with the bases are pronounced by moving the tongue toward the bases.

However, unlike Korean, English is not used as a consonant. In other words, the vowels move toward the consonants and do not pronounce. If the consonant is used as a support, a gap is created between the consonant and the consonant in the bundle of consonants. This is one of the most important differences between the pronunciation of Korean and English.

English also has a vowel pronunciation that moves the tongue. This is called a middle vowel (double vowel or triple vowel). It is a pronunciation that moves the tongue toward the subordinate vowel, starting with the main main vowel. Therefore, the English vowels can be thought of as having the main vowels as the supporting vowels.

Consonant Independent Pronunciation Law or Consonant Self Pronunciation Law

There is no vowel pronunciation between the consonant and the consonant in a bunch of consonants. If vowels are added between consonants and consonants, the number of syllables increases. If the number of syllables is different, the other person will not understand at all.

Therefore, except for the consonants immediately preceding the vowels, all consonants are pronunciations that are not directly connected to the vowels, that is, independent vowels without vowels.

Economic Law of Pronunciation

Move from one consonant to the next shortest. Thus, all consonants are pronounced extremely short except for the last consonant of the consonant bundle. When the same consonants overlap, they are pronounced only once. However, the rhythm must be pronounced separately (eg orange juice).

Non-rupture law of rupture

Immediately after the rupture sound is another ruptured, rubbing or stinging sound. In Korean, when a ruptured sound comes to the base, it is pronounced non-ruptured. As the support is important in Korean, non-ruptured pronunciation is very important in English, and if omitted, it is very awkward.

However, when the ruptured sound is the first word of a bunch of words, it is usually pronounced strong bursting. In addition, all consonants should be pronounced except in exceptional and abbreviated forms. It is very different from omitting the pronunciation and not hearing it weakly. (E.g. in roast beef, t is pronounced non-ruptured)

In the case of (strong vowel + nt + weak vowel), it is a tendency to omit 't' (e.g. center, county).

(Law of the weak 1)

In particular, this last consonant bundle of words ending with a consonant cluster is weakly pronounced because the vowels do not support it from behind. In some cases, the sound is so weak that it is almost inaudible. But be sure to pronounce. Therefore, if the end of the word is broken sound pronounced weak burst.

(Law 2 of the abbreviation)

The pronunciation of a weak consonant bundle does not change to a strong pronunciation even if there is a strong accent in the first vowel of subsequent words. In other words, the abbreviation is an eternal abbreviation.

An example of a consonant pronunciation is as follows.

TR [tr] (e.g. tree)

FIG. 15A illustrates the movement of the tongue when pronounced 't', and FIG. 15B illustrates the movement of the tongue when pronounced 'tr'.

When you look at the movement of the tip of the tongue during the pronunciation of 't', it is pronounced by falling down from the upper root as shown in Figure 15a. However, if 't' is followed by 'r', the tip of the tongue is moved to the pronunciation position of r near the gum as shown in Figure 15b by the economic principle of pronunciation.

At this time, due to the c's sound (wind following the sound of the rupture sound), a short friction sound is heard between the tip of the tongue and the gum. In other words, t is not deliberately pronounced in the form of a palpation, but is a result of incidentality caused by the direction of the tip of the tongue.

s + burst (sp, sk, st)

In order to pronounce strong burst sounds, instantaneous air compression is required. It is very hard to blow the air quickly to pronounce the friction sound 's' and then pronounce the momentary burst. Therefore, weak burst pronunciation is made. (E.g. last, grasp, desk)

Weak burst pronunciation is a sound that bursts quickly by touching the tongue very lightly in the same position as a strong burst. At this time, if the voiced sound is followed by the sound of the weak burst, the voice changes to the voiced burst. Sticking the tongue very lightly on the gums Rapidly pronounced k, p, t sounding ruptures are strongly different from our hard (ㄲ, ㅃ, ㄸ) pronounced rupture. (E.g. sky, spy, stay)

Pronunciation of the word bunch

When we speak, we cut and pronounce each word or pronounce several words together rather than tying the whole sentence together. This unit, which connects several words and pronounces it like a word, is called a word bundle.

The pronunciation law of consonants is also applied to the pronunciation of a word bundle. However, the last consonant of a word that ends with a consonant is weakly pronounced even in the middle of the word. In other words, even if a strong vowel word is connected immediately after, the weak pronunciation of the previous word is not strong.

E.g. sports, sports events

In the case of Sports, t is pronounced weak, non-ruptured. However, in the case of sports events, voiced sounds are followed by voiced non-ruptures.

Hangul notation of English

There are more than 40 phonemes (consonants and vowels) in English. Only a few of them exactly match the Korean phonemes.

However, Korean consonants and vowels of English are needed to be able to write as close to the original sound as possible using similar Korean and its variations, and to be able to guess the original English relatively accurately. Note that the use of Hangul notation, the pronunciation is significantly different from the original Hangul.

Consonant notation

16 is a consonant table written according to a notation method according to an embodiment of the present invention.

As can be seen from the table, every consonant uses a suffix font when it has a vowel or Korean vowel.

(earth)Example:

(earth)

Vowel notation

17 is a vowel table marked according to a notation method according to an embodiment of the present invention.

r'의 경우 단모음이므로 입과 혀를 움직이지 않고, 'ar', '

r'의 경우 r뒤에 자음이 오거나 r로 끝나는 경우 r을 종속 모음으로 가지는 중모음으로 발음한다.'

r 'is a short vowel, so the mouth and tongue do not move,

In the case of r ', if a consonant follows r or ends with r, it is pronounced as a vowel with r as a dependent vowel.

'의 발음은 '

r'의 발음보다 입을 크게 벌리기 때문에 혀만 움직여 발음이 가능하나 'I

r', '

r', 'u

r'의 경우 입을 더 벌리면서 발음한다.Also, 'a' and '

'Is a pronunciation of'

Because you open your mouth wider than r 'pronunciation, you can pronounce it by moving your tongue.

r ','

r ',' u

In the case of r 'it pronounces with its mouth wider.

In the case of writing English vowels in Korean, dependent vowels can be written in a form that is distinct from the main vowel, for example, superscript or subscript.

Consonant wad notation

In English, consecutive consonants are replaced with corresponding consonants in Korean. Accordingly, it is possible to input only the consonants of three or more consecutive consonants, and to input consonant bundles as in English using Korean.

Example) la ㅅㅌ _the blah (strike), oi _a directory _(story), below ㅅㅌ (last), in ㅅㅌ (lost), ㅅpo _d ㅌㅅ (sports), US ㄹㅋ (milk)

According to the present invention, it is possible to synthesize English voices closer to the pronunciation of native speakers using Hangul.

The present invention has revealed important structural differences between Korean and English pronunciations, and provides rules that can be converted from native English pronunciations to native pronunciations.

It also created a new English Korean notation, opening the way for unification of pronunciation education.

By using the laws and notation of the present invention, it is possible to learn the pronunciation of native speakers, and it will greatly help Korea to enter bilingual culture.

Although the present invention has been described in terms of some preferred embodiments, the scope of the present invention should not be limited thereby, but should be construed as a modification or improvement of the above embodiments supported by the claims.

Claims (4)

A Hangul input unit for receiving Hangul input; A voice converter for converting the Hangul input into a predetermined English voice; And A speech synthesis apparatus including an English speech output unit for outputting the English speech, The Hangul input unit is an English speech synthesis device using Hangul, characterized in that for performing the initial input formed of three or more consonants. The method of claim 1, The Hangul input unit is an English speech synthesis apparatus using Hangul, characterized in that for performing the input of the dependent vowel distinguished from the main vowel. A Hangul input step of receiving Hangul input; A voice conversion step of converting the Hangul input into a predetermined English voice; And A voice synthesis method comprising the step of outputting the English voice, the English voice output step, The Hangul input step is an English speech synthesis method using Hangul, characterized in that for performing the initial input formed of three or more consonants. The method of claim 3, wherein The Hangul input step is an English speech synthesis method using Hangul, characterized in that the input of the dependent vowel distinguished from the main vowel.

Images