JPS6111021A - Speaking exercise apparatus - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a speech training device for speech-impaired persons, and more particularly to a speech training device that uses not only sound waves but also speech information from other speech organs.

Structure of the conventional example and its problems It is estimated that there are several hundred thousand people with language disabilities in Japan who are unable to speak due to hearing impairment, etc., but there is a shortage of language trainers,
It is difficult to say that sufficient training time is secured due to the fact that training takes a long time. Although it is desired to develop an effective speech training device, a training device that can classify and recognize speech sounds and display whether the speech is correct or incorrect has not yet been obtained.

As a device closely related to speech training devices, there is a speech recognition device, which has been actively researched and developed in recent years. Most normal speech recognition devices use only information from speech waves, but from only speech waves, it is difficult to recognize the consonants /p, t, /, /b, d.
, g/ and 7m, n.

It is difficult to accurately detect and distinguish between each other.

Therefore, in addition to voice waves, there are vocal cord vibrations and nasal vibrations.

A pronunciation feature extraction device (see Japanese Unexamined Patent Publication No. 160997/1983) that uses information such as air flow and presence or absence of contact between the tongue and the palate has been proposed.

FIG. 1 shows criteria for determining whether to perform phoneme classification in the conventional pronunciation feature extraction device. That is, in Fig. 1, the voiced sound / b+ d+ g y ” vn, y7
/ has vocal cord vibration (→, and is a voiceless sound /p.

For t, k, and h/, there is no vocal cord vibration (c). Nasal/m.

n, y)/ has nasal vibration (a), /p, t
, k, -b, d, g, h/ have no nasal vibration. The plosives /p, t, b, d, g/ and the fricative /h/ have a velocity of oral airflow (g), but the velocity of 7m, n, v/ has none (c). Furthermore, the rate of change of the oral airflow velocity is /p,t.

b, 4/ is large and becomes true (g), but /k, g
, h/small and absent ←). Regarding palatal contact,
/p.

b, m, h/ are not closed patterns, /1°d r
In /k, there is a pattern of prepaleoclosure.

g and n/ result in a postfix closure pattern. Therefore, the first
It is assumed that phonemes can be classified according to the diagram.

However, according to experiments, with the above configuration, (1) voiced plosives /b, d, g/ have a strong tendency to become nasalized, and nasal vibrations may become present (a).

(2)/ni, g,? Posterior closure in J/ may not be detected at all due to individual differences, and in this case no closure occurs.

(3) The articulatory point of rupture is in the posterior tongue /k, g/,
The detection of the flow velocity of the oral airflow is somewhat unstable compared to /p, t, b/, and [flow velocity, rate of change] is [Yes (→, Yes (+))]
, C presence (→, nothing (c)], [no←), nothing←)] may vary. A similar tendency is also seen for /d/, although it is less frequent.

Due to these causes, the criteria shown in Figure 1 is /p/ and /, /C1/ and /n/.

It has been found that it is difficult to use it as is for speech training because confusion arises between /g/ and /ri/ and /m/.

OBJECTS OF THE INVENTION An object of the present invention is to provide a speech training device that can classify and recognize speech sounds and display whether they are correct or incorrect, in accordance with actual speech training, by solving the difficulties in applying the conventional speech recognition device. do.

Structure of the Invention The present invention is a speech training device that is equipped with a detector for information on voice waves, nasal vibrations, laryngeal vibrations, and closed contact between the palate and the tongue, a classification recognition section, and a training sound designation section. By performing classification recognition focusing only on phonemes that are similar to the designated sound in terms of phonetic production, difficulties in speech recognition can be eliminated and classification recognition that is suitable for actual training can be realized.

DESCRIPTION OF EMBODIMENTS FIG. 2 is a block diagram showing an embodiment of the present invention5.
In Figure 2, 1 is an audio wave detector used to obtain information on the presence or absence of sound and the beginning of the sound, and 2 is installed near the center of Makabe.
A nasal vibration voice detector is used to obtain information about nasal sounds. 3 is a laryngeal vibration detector that is not attached near the vocal cords of the larynx and is used to obtain information about unvoiced sounds. 4 is an oral airflow detector placed in front of the oral cavity to detect changes in flow velocity, for example. 5 is a tongue rupture detector that obtains rupture information by detecting the rate of rupture; and 5 a tongue rupture detector that obtains closure contact information from a palate contact detector attached to the palate in the oral cavity to detect contact between the tongue and the palate. A closure detector, 6 a classification recognition unit that classifies and recognizes speech sounds from the detection information of 1 to 6 above, 7 a utterance correctness determination result display unit that shows the results of the classification recognition unit, 8 a training prior to VLL training. A training sound specifying section 9 receives and accepts the designation of sounds, and a control section 9 controls the classification recognition section 6, the correct/incorrect determination result display section T, and the training sound specifying section 8.

The operation of the classification recognition unit 6 in the embodiment of the present invention configured as described above will be described in detail below according to an example.

Speech training does not involve practicing various phonemes in a chaotic manner, but is carried out systematically, taking into account similarities in phonemes, opposition, and two levels of difficulty. FIG. 3 shows an example of the Pabama table, which is the most basic and important training goal in speech training. Training begins with /p/ and moves to adjacent sounds in rows and columns as the user becomes proficient.

FIG. 3 has common features depending on the rows and columns. That is, from the point of articulatory position, /p, b, m/ are bilabial sounds, /t, 6, n/ are alveolar sounds, /ni, g,
n/ is a soft palate sound. In terms of the type of sound, /p,
t.

k/ is a voiceless plosive; b, d, g/ are voiced plosives.

7 m+ n+')/ is a nasal sound. As can be inferred from these points, even in actual speech training, confusion tends to occur between sounds that are adjacent in the row and column directions, and confusion between sounds that are far apart is rare. For example, when uttering /p/, it may be mistaken for /1/ or /b/, but it is very rare for it to be mistaken for /ni/, /m/, or other sounds. In other words, according to the actual training, when training /p/, /p/ (correct answer), /l
Still b/ (mistake to adjacent sound), /? / (other mistakes) classification recognition, or /p/, /l/, /
b/.

By performing the classification recognition of /9/, it is possible to carry out sufficiently effective training without performing the difficult classification recognition of /1)/ and /ni/.

Similarly, if you know the speech sounds to be trained, /g/ and /m
Training is possible without performing classification recognition of /, /)7/ and /m/.

Next, /d and g/ have a strong tendency to become nasalized, sometimes exhibiting nasal vibrations to the same extent as nasal sounds, and have been the cause of interlude, but according to experiments, /d and g/ have become nasalized. It was found that for g/, strong nasal vibrations rarely last for more than 100 msec, and on the contrary, for nasal sounds, strong nasal vibrations last for more than 120 msec in most cases. Therefore, by utilizing this fact, it is possible to almost eliminate confusion between /+1/ and /n/, /g/ and /ri/.

Based on the above idea, FIG. 4 shows the types of determination results and combinations of separation functions used for determination when classification recognition is performed for each sound training shown in FIG. 3.

In the training of /1/, /p/ and /ni/ were not classified/recognized.
, k / is used as the judgment result. This is because, as mentioned above, training can be carried out even if / p, k / are grouped together as an error for adjacent sound -1, and from a training standpoint, it is better not to make incorrect correct/incorrect judgments even if the performance of classification recognition is degraded. This is because it is desirable and the classification recognition unit can be realized simply. /b, g/, / in the training of /d/
The same goes for 7m and 77/ in the training of n/.

Figure 5 illustrates the criteria for realizing each separation function in Figure 4. For example, the voiceless plosive /p/ has no nasal vibration, is plosive, is voiceless, and has a tongue This indicates that there is no closed contact.

Tongue closed contact information is used as the separation condition between columns. That is / P
+ br m/ and /l, d, n/ are separated by the absence and presence of tongue-closed contact, /t, d,
The criterion for separating n/ from /k, g, and ri/ is the presence or absence of tongue-closing photoptera. /p, t, k/
The criteria for separating the line spacing between /b, d, and g/ are unvoiced and voiced.

/b, d, g/ and 7m, n, 7J/ (The conditions for determining the difference between D lines are a tendency to nasalization and strong and long nasal vibrations.

In addition, the separation function uses the characteristics that both have in common as a common basic condition, and if the common basic condition is not satisfied, the classification recognition is "?"
”(other).

The sixth chrysanthemum is based on FIG. 6, with the common basic conditions and separation judgment conditions relaxed for each separation function.

The operation of the speech training device of this embodiment, which is equipped with the classification recognition unit configured as above, will be explained below using a flowchart.

91] in the case of /p/ training is shown in FIG.

This will be explained using FIGS. 2 and 7.

(B) Before starting the training without switching, use the training sound specified sound b8.
Specify which sound to train. If /p/ is specified, the classification recognition unit 6 is notified via the control unit 9 that the training for this/ra/p/ will start.

(b) When the next training sound is uttered, the classification recognition unit 6
Now, information from detectors 1 to 6 is received and classification recognition of /p/ is started.

In the classification recognition of 1'e/p/, first, a separation function for p/ is activated, and the basic conditions and separation judgment conditions shown in FIG. 6 are checked. If a basic condition is violated, the violation condition is maintained and the process proceeds to p/b separation).

Next, the separation judgment condition is checked, and if there is a tongue-closed tangent worm 75, it is judged as /1/, and the judgment sound /1/ is displayed on the success/failure judgment result display section 7 through the control section 9.
Reason 75 for this determination is shown below.

) In the p/b separation function, Article lf+75 in Figure 6
Two can be investigated. If the basic conditions are violated, keep the violation conditions and proceed to (). In the separation judgment condition, the presence/absence of voicing is checked, and if it is voiced, it is determined as /b/, and the judgment sound /b/ and the reason for the judgment are displayed via the control unit 9 on the correct/incorrect judgment result display unit 7. is shown.

(e) It is checked whether there are any violation conditions using the p/ separation function, and if there is a violation, proceed to (v). If there is no violation, /p/, which is the correct answer, is displayed on the correctness determination result display section 7.

(f) This processing step violates the basic condition. The correct/incorrect judgment result display section 7 shows /? / indicates the content of the violation. The separation functions activated in the training of /p/ are the p/ separation function and the p/b separation function, and violations of the fundamental conditions of these separation functions are ``pion u, u failure''.

There are four types: ``Yusei'' and ``Shitaheisa Ali''.

FIG. 8 shows the types of messages that violate the basic conditions for each training sound.

Effects of the Invention The speech training device of the present invention uses voice waves, nasal vibrations, laryngeal vibrations,
It is equipped with a detector that obtains information on the presence or absence of closed contact between the palate and the tongue, a classification recognition section, a training sound designation section, a correct/incorrect judgment result display section, and a control section. By specifying a sound and performing classification recognition focusing on the training sound, it is possible to classify and recognize speech sounds and display whether the recognition is correct or not, which is extremely effective in speech training.

[Brief explanation of the drawing]

Fig. 1 is a diagram showing criteria for determining whether or not to perform phoneme classification in a conventional pronunciation feature extraction device, Fig. 2 is a block diagram showing a speech training device according to an embodiment of the present invention, and Fig. 3 is a diagram of a Pabama table. An array diagram showing an example, FIG. 4 is a correspondence diagram showing the types of judgment results for each training sound and the types of separation functions used to obtain the judgments in the embodiment of the present invention, and FIG. FIG. 6 is a diagram illustrating common basic conditions and separation determination conditions based on FIG. 6, and FIG. 7 is a flowchart showing an example of the operation of this embodiment.
FIG. 8 is a correspondence diagram showing the types of violation messages for each training sound. 1... Sound wave detector, 2... Nasal vibration detector, 3... Laryngeal vibration detector, 4... Rupturability detector, 6...・・Tongue closure detector, 6・・・・・・
Classification recognition unit, 7...Accuracy determination result display unit, 8.
. . . Training sound designation section, 9 . . . Control section. Patent applicant: Director of the Agency of Industrial Science and Technology Kawa 1) Hirobe Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8

Claims (1)

[Claims] A sound wave detector, a nasal vibration detector, a laryngeal vibration detector, a burst detector, a tongue closure detector that detects closure contact information between the tongue and the hard palate, and speech sound classification. a classification recognition section for performing the training sound, a training sound specification section for specifying the training sound, a correctness judgment result display section for indicating the result of the classification recognition, the classification recognition section, the training sound specification section, and the correctness judgment result display section. What is claimed is: 1. A speech training device comprising: a control unit for controlling the speech training device; the speech training device is characterized in that a training sound is specified before training starts, and classification recognition is performed focusing on the training sound;