JPS6331795B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a plosive sound recognition device for recognizing plosive sounds.

In recent years, research into voice recognition devices has become active, and some devices have been put into practical use and are now on the market. Expectations are also growing that it will become the most natural means of communication between computers and humans, which continue to advance rapidly.

However, a device that can correctly recognize the utterances of any speaker has not been obtained, and it is particularly difficult to detect and classify irregular consonants, and it is also very difficult to reliably detect and classify plosive consonants.

Furthermore, the conventional method requires a large number of arithmetic operations including multiplication and division, such as calculating a feature vector from the outputs of a large number of band filters and calculating frequency bias.

An object of the present invention is to provide a plosive recognition device that reliably detects and classifies plosive consonants of any speaker. Another object of the present invention is to provide a plosive recognition device that does not require multiplication and division in the plosive recognition unit and has simple arithmetic processing.

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of a plosive recognition device in one embodiment of the present invention. In the figure, reference numeral 1 denotes a voice wave detector, such as a low-noise close-talk type microphone. Reference numeral 2 denotes an exhalation flow rate detector, for example, a hot wire flowmeter sensor, etc., which is placed in front of the oral cavity to detect the exhalation flow rate. Reference numeral 3 denotes a laryngeal vibration detector, such as a vibration pickup, which is attached to the larynx near the vocal cords using medical double-sided tape or the like to detect laryngeal vibration. 4 is a palate contact detector that detects contact information between the tongue and the hard palate, and an example of its shape is shown in FIG. The palate contact detector 4 has a plurality of electrodes 4a arranged, and when used, it can detect the state of contact between the tongue and the hard palate by attaching it to the hard palate in the oral cavity. FIGS. 3A to 3D are schematic patterns showing the state of contact between the tongue and the hard palate, and the hatched areas indicate the contact areas. Figure a shows a pattern typically seen in /S, Z/ utterances, and figure b shows /i, j,
Patterns seen when making utterances such as ∫／, Figure c
is a pattern typically seen when uttering /t, d, n/, etc., and d in the figure is a pattern typically seen when uttering /r/.

Reference numeral 5 denotes a voice wave intensity detector, for example a detection smoothing circuit, which extracts the envelope of the voice wave. Reference numeral 6 denotes a laryngeal vibration intensity detector, which is, for example, a detection smoothing circuit, and extracts the envelope of laryngeal vibration. Reference numeral 7 denotes a temporary storage unit for detection information, which stores audio wave intensity information that is the output of the intensity detector 5, expiratory flow rate information that is the output of the expiratory flow rate detector 2, and laryngeal vibration intensity information that is the output of the intensity detector 6. and palate contact information, which is the output of the palate contact detector 4, are temporarily stored. 8 is a plosive sound recognition unit that recognizes plosive sounds based on information in the temporary storage unit;
A more detailed explanation will be given below using FIG. 4.

In FIG. 4, reference numeral 81 denotes a sound section search section, which searches for a section with sound based on the audio wave intensity information A in the temporary storage section 7. Reference numeral 82 denotes a plosive sound inspection section which, when a section with a sound is found, tests whether it is a plosive sound based on the expiratory flow velocity information in the temporary storage section 7. Reference numeral 83 denotes a contact pattern inspection unit, which performs a classification inspection on /t, d/ and /K, P, g, b/ among plosives based on the palate contact information d in the temporary storage unit 7. Reference numeral 84 denotes a first voiced sound testing section, which performs / based on the laryngeal vibration intensity information in the temporary storage section 7.
t and d/ are inspected, and /t/ and /d/ are recognized and classified. 85 is a second voiced sound test section, which detects /k, p, g, b/ based on the laryngeal vibration intensity information section c.
is inspected and classified into two categories: /k,p/ and /g,b/. Reference numeral 86 denotes a first expiratory flow rate testing section, which tests /k and p/ based on the expiratory flow rate information, and recognizes and classifies /k/ and /p/. 87 is the second
The expiratory flow rate testing unit tests /g and b/ based on the expiratory flow rate information, and recognizes and classifies /g/ and /b/.

The phonemes uttered by the plosive recognition device configured as described above are /t/, /d/, /k/, /
It is possible to recognize which of p/, /g/, and /b/ is a plosive, and whether it is silent or non-plosive.

The operation of the plosive recognition unit 8 will be described below with reference to the flowchart shown in FIG.

(a) First, search for a voiced section. According to the audio wave intensity information in the temporary storage unit 7, if the audio wave intensity is larger than the threshold value determined through experiments and satisfies the conditions for duration length, for example, 80 msec or more, the section is determined to be a sound section. do. If the above conditions are not met, it is recognized as silence.

(Sound section search unit 81) (b) Check whether the sound is a plosive or not. According to the expiratory flow velocity information in the temporary storage unit 7, before and after the start of the sound interval, there exists an expiratory flow interval in which the expiratory flow velocity satisfies the conditions for the threshold value and duration length determined by experiment, and the expiratory flow interval is at the beginning. When the rate of change in expiratory flow rate is larger than a threshold determined experimentally, it is recognized as a plosive sound, and in other cases, it is recognized as a non-plosive sound.

(Plosive sound inspection section 82) (c) Next, plosive sounds /t, d/ and /k, p, g,
Classified into two categories: b/. In front of the beginning of the sound section, the palate contact information in the temporary storage section 7 is checked, and it is determined that there is a closing contact pattern TDN or R (see Figure 3 c, d) and that the duration is determined by the experiment. If the condition is satisfied, /t,
d/If any of the conditions are not satisfied/
Let it be one of k, p, g, b/.

(Contact pattern inspection department) (d) If it is either /t or d/, /
Recognize t/ and /d/. Before and after the start of a voiced section, a section in which the laryngeal vibration intensity satisfies the conditions for the threshold value and duration length determined through experiment is checked based on the laryngeal vibration intensity information in the temporary storage section 7, and that section is designated as a voiced section. shall be.

Find the time difference between the start of the voiced section and the start of the voiced section, and if the start of the voiced section is ahead of the start of the voiced section by a certain amount of time determined by experiment, for example 20 msec or more, the sound is voiced.
Others are determined to be voiceless sounds. Therefore, voiced plosive/
It is possible to recognize and classify d/ and voiceless plosive /t/.

(Voiced Sound Inspection Unit 84) (e) If it is any of /k, p, g, b/, it is classified into voiced sound /g, b/ and unvoiced sound /k, p/. Determination of voiced and unvoiced sounds is described in (d) above.
This is performed in the same manner as the voiced sound testing section 84.

(Voiced sound test unit 85) (f) If it is either /k or p/, /k/
Recognizing /p/, with the labial plosive /p/,
The beginning of the sound section and the beginning of the expiratory flow section are observed almost simultaneously, whereas the deep velar plosive sound/
In k/, the expiratory flow section is observed after the sound section. Compare the start of the sound section and the start of the expiratory flow section, and if the start of the sound section precedes the start of the expiration flow section by a certain amount of time determined by experiment, /k/, etc. In the case of /p/
I recognize that.

(Expiratory flow rate test section 86) (g) If either /g or b/, /g/
Recognize and /b/. The relationship between the start of the voiced section and the start of the expiratory flow section in the labial plosive /b/ and the velar plosive /g/ is the same as that for /p/ and /k/ in f. Compare the start of the sound section and the start of the expiratory flow section, and if the start of the sound section precedes the start of the expiration flow section by a certain amount of time determined by experiment, /g/, etc. In this case, it is recognized as /b/.

(Expiratory flow rate testing unit 87) As described above, according to this embodiment, a sound interval is searched based on the sound wave intensity information, a plosive is determined based on the expiratory flow rate information, and a plosive is determined based on the palate contact information. plosives /t, d/ and /k, p, g,
After classifying into b/, voiced plosives /d,
Classify g, b/ and voiceless plosives /t, k, p/. Furthermore, by separating the labial plosive sounds /p, b/ and the plosive sounds /k, g/ near the back tongue from the phase difference between the expiratory flow section and the sound section, which were examined based on the expiratory flow velocity information. , the uttered phoneme is a plosive /t/, /d/, /k/, /p/, /g/, /
It is possible to recognize whether it is b/, silent or non-plosive.

In this example, /t, d/ and /k, p, g,
Although voiced plosives and voiceless plosives are classified after performing the two classifications of b/, contact patterns may also be examined after classifying voiced sounds and unvoiced sounds. It is also possible to configure the sound interval search section 81 and each of the inspection sections 82 to 87 to perform test classification without determining the conditions for each duration length, and finally to determine the conditions for the duration length. can.

As described above, the present invention is configured to temporarily store voice intensity information, expiratory flow velocity information, laryngeal vibration intensity information, and palate contact information, and recognize plosive sounds based on this temporarily stored information. plosive consonants can be reliably detected and classified.
Furthermore, it is possible to realize a plosive sound recognition device that requires no multiplication or division in the plosive sound recognition unit and has simple arithmetic processing, which has a significant effect on speech recognition and the like.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a plosive sound recognition device according to an embodiment of the present invention, FIG. 2 is a diagram showing an example of the shape of a palate contact detector, and FIGS. FIG. 4 is a block diagram showing a specific example of the plosive recognition unit, and FIG. 5 is a flowchart for explaining the operation of the plosive recognition unit. 1...Audio detector, 2...Expiratory flow rate detector, 3
... Laryngeal vibration detector, 4 ... Palate contact detector, 5
...Intensity detector, 6...Intensity detector, 7... Temporary storage section, 8... Plosive sound recognition section.

Claims (1)

[Claims] 1. Means for detecting voice waves, means for obtaining voice intensity information from the detected voice waves, means for detecting expiratory flow velocity, means for detecting laryngeal vibrations, and means for obtaining laryngeal vibration intensity information from the detected laryngeal vibrations. and a means for detecting contact information between the tongue and the hard palate.
a temporary storage section that temporarily stores the voice intensity information, expiratory flow rate information, laryngeal vibration intensity information, and palate contact information; , g) is extracted and separated into (t, d) and (k, p, g, b) based on palate contact information, and (t), (d), (k, p), (g, b), and further includes a plosive sound recognition unit that separates and identifies (k), (p), (g), and (b) based on sound wave intensity information and expiratory flow rate information. A plosive sound recognition device characterized by: