JPH036519B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to a pronunciation feature extraction device that recognizes pronunciation from information other than speech. Speech is converted into voice when the exhaled airflow from the lungs passes through the vocal cords in the larynx, which vibrate, and is modulated by changing the shape of the exhaled air passage leading to the lips and nasal cavity. The first cry is produced as a result of the comprehensive function of the ducts. Conventionally, to extract such sounds, the sound waves are converted into electrical signals using an acoustic microphone.
The output is sent to a large number of filter circuits having a predetermined frequency band, and the pronunciation is characterized based on the output of each filter circuit. However, it is extremely difficult to perform speech recognition by extracting the pronunciation characteristics of all phonemes using only speech waves, which is the result of the comprehensive functioning of the vocal organ. In particular, non-stationary consonants have strong noise energy, and with the exception of voiceless fricatives |S, ∫|, etc., whose features can be almost certainly extracted from the speech wave, voiceless fricatives |h| and voiceless plosives |p, t , k|, voiced plosives |b, d, g|, nasal sounds |m, n, 〓|, etc. are extremely difficult to detect and separate. In view of the above-mentioned drawbacks, the present invention has been proposed by installing or arranging a detector for detecting the movement of each part of the voice tube near each part of the voice tube, and by having the output from each of the detectors processed by a processing device. The present invention also provides a pronunciation feature extraction device that can accurately extract pronunciations. An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a block configuration of a pronunciation extracting device according to an embodiment of the present invention. In the figure, 1 is a vocal cord vibration detector that is attached near the vocal cords of the larynx and detects vibrations of the vocal cords, 2 is a nasal airflow detector that is placed in front of the nasal cavity and detects nasal airflow, and 3 is a nasal airflow detector that is placed in front of the oral cavity and detects oral airflow. 4 is a palate contact detector that is attached to the palate in the oral cavity and detects contact between the tongue and the palate. Reference numeral 5 denotes a processing device for extracting pronunciation features from the outputs of the vocal fold vibration detector 1, nasal airflow detector 2, oral airflow detector 3, and palate contact detector 4. A detailed explanation will be given below. In FIG. 2, 6 is a threshold circuit that determines the presence or absence of vocal fold vibration based on a specific value from the vocal fold vibration information of the vocal fold vibration detector 1, and 7 is a threshold circuit that determines the presence or absence of vocal fold vibration based on a specific value from the nasal airflow information of the nasal airflow detector 2. A threshold circuit 8 determines the rate of change (acceleration) of the oral airflow by differentiating the oral airflow information from the oral airflow detector 3; 9 a differentiation circuit that determines the presence or absence of the rate of change of the oral airflow; Threshold circuit that determines based on a specific value, 1
0 is a threshold circuit 11 that determines the presence or absence of oral airflow based on a specific value from the oral airflow information of the oral airflow detector 3;
The palate contact information from the palate contact detector 4 is once converted into a contact signal between the tongue and the palate by the measurement circuit 12, and then the three signals of anterior tongue closure, posterior tongue closure, and no closure described below are detected.
13 is a phoneme classification circuit that performs phoneme classification based on the presence or absence of each threshold information output from the threshold circuits 7, 7, 9, and 10, and three types of information in the tongue closure detection circuit 11; It is a circuit. A specific method of using the pronunciation feature extracting device configured as described above will be explained below with reference to FIG. As shown in FIG. 3, the vocal cord vibration detecting section 1 detects vocal cord vibration by attaching an acceleration sensor 1' to the vocal cord part of the larynx of the human body with medical double-sided tape. The detected vocal fold vibration is output to the threshold circuit 6, which sends a positive (+) signal to the phoneme classification circuit 13 if the value of the vocal fold vibration is above a certain value, and sends an absent (+) signal to the phoneme classification circuit 13 if it is below a certain value. −) Output a signal. Nasal airflow is detected by attaching a hot wire current meter sensor 2' as the nasal airflow detector 2 to the tip of a movable arm fixed to the pivot portion of the headband and placing it in front of the nasal cavity of the human body. The detected nasal airflow is output to the threshold circuit 7,
The threshold circuit 7 outputs a presence (+) signal to the phoneme classification circuit 13 if the value of the nasal airflow is above a specific value, and outputs an absence (-) signal if it is below a certain value. Oral airflow is detected by fixing and arranging a hot wire flow meter sensor 3' as the oral airflow detector 3 on a desk or the like in front of the oral cavity of the human body. The detected oral airflow is output to a differentiating circuit 8, which determines the rate of change in the oral airflow and outputs the rate of change to a threshold circuit 9. Then, the threshold circuit 9 sends a positive (+) signal to the phoneme classification circuit 13 if the rate of change value is above a specific value, and a negative signal (-) if the rate of change is below a certain value.
Output a signal. On the other hand, the oral airflow detected by the hot wire flow meter sensor 3' is also output to the threshold circuit 10, and if the value of the oral airflow is above a specific value, the threshold circuit 10 sends a positive (+) signal to the phoneme classification circuit 13.
Moreover, if it is below a certain value, a null (-) signal is output. Further, as the palate contact detector 4, a contact sensor 4' as shown in FIG. 4 is used. The contact sensor 4' has a large number of electrodes 4'a on the part that comes into contact with the tongue, is attached to the roof of the oral cavity in the human body by a stopper part 4'b, and detects the state of contact with the tongue by the electrode 4'a. . The detected contact state between the electrode 4'a and the tongue is sequentially input to the measurement circuit 12 and the tongue closure detection circuit 11, and when the contact state becomes a pattern as shown in Fig. 5A, it is determined that the anterior tongue is closed. When the information becomes a pattern as shown in FIG. Finally, in the phoneme classification circuit 13, threshold circuits 6, 7,
Speech can be determined based on each piece of information input from 9, 10 and the tongue closure detection circuit 11.

【table】

[Table] For example, when the voice "hana" having a phoneme wave as shown in FIG. 6A is uttered, the acceleration sensor 1' outputs a waveform as shown in FIG. 6B to the threshold circuit 6. Then, in the threshold circuit 6, judging from a specific threshold value, there is no (-) signal at the "h" part,
In the “n” part, the presence (+) signal is sent to the phoneme classification circuit 1.
Output to 3. Further, the hot wire flowmeter sensor 2' outputs a waveform as shown in FIG. 6C to the threshold circuit 7. Then, the threshold circuit 7 outputs an absent (-) signal for the "h" portion and a present (+) signal for the "n" portion to the phoneme classification circuit 13, based on a judgment based on a specific threshold value. Furthermore, the hot wire flow meter sensor 3' outputs a waveform as shown in FIG. Then, in the threshold circuit 9, the differential value from the differentiating circuit 8 is judged from a specific threshold value, and it is determined as "h" and "n".
A null (-) signal is output to the phoneme classification circuit 13 at the part. The threshold circuit 10 also outputs a presence (+) signal at the "h" portion and a no (-) signal at the "n" portion to the phoneme classification circuit 13 based on a judgment based on a specific threshold value. On the other hand, the contact sensor 4' detects the contact state between the electrode 4a and the tongue, and outputs the detected state to the tongue closure detection circuit 11 via the measurement circuit 12. and tongue closure detection circuit 11
outputs to the phoneme classification circuit 13 information of "no closure" in the "h" part and information of "frontal tongue closure" in the "n" part, depending on the contact pattern. Then, the phoneme classification circuit 13 selects "h" from an internal memory table as shown in the table based on each information.
and "n" can be recognized. As described above, the movement of each vocal tube is detected by the vocal fold vibration detector 1, the nasal airflow detector 2, the oral airflow detector 3, and the palate contact detector 4, and the information detected by each detector is processed by the processing device 5. By selecting and determining a specific phoneme from the width of a pre-stored table based on the width of the table, it is possible to accurately recognize speech, which has been difficult in the past. As described above, the present invention combines vocal cord vibration information detected by a vocal cord vibration detector, airflow information in front of the nasal cavity detected by a nasal airflow detector, oral airflow information detected by an oral airflow detector, and palate contact detector. Each phoneme of plosives and nasals can be identified more accurately than before based on the contact information between the tongue and the roof of the mouth detected by the system, and its practical effects are significant.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a pronunciation feature extraction device according to an embodiment of the present invention, FIG. 2 is a block diagram of processing in the same pronunciation feature extraction device, and FIG. 3 is a diagram showing an example of use of the same pronunciation feature extraction device. FIG. 4 is a plan view of the contact sensor, FIG. 5 is a diagram showing a contact pattern between the tongue and the palate, and FIG. 6 is a waveform diagram of each detector. 1... Vocal cord vibration detector, 2... Nasal airflow detector,
3... Oral air flow detector, 4... Palate contact detector, 5
...Processing device.

Claims (1)

[Claims] 1 Equipped with a vocal fold vibration detector attached to the larynx, a nasal airflow detector placed in front of the nasal cavity, an oral airflow detector placed in the front of the oral cavity, and a palate contact detector that detects contact between the tongue and the palate. , extracting the grooves of the plosive sounds p, t, k, b, d, g, and h based on the output of the oral airflow detector, and extracting the nasal sounds m, n, 〓 based on the output of the nasal airflow detector, Based on the output of the vocal cord vibration detector, p, t, k, h and b, d, g
and p, h, t, k, b, d, g, m, n, 〓 based on the output of the palate contact detector. 1. A pronunciation feature extraction device comprising: a processing device that separates p and h based on a rate of change in airflow.