JPS617868A - Enunciation training 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 Industrial Application] The present invention relates to a vocal training device that determines and evaluates the stability of the vocalizations of four vocal trainees and displays the results in one form.

Vocal training devices used for language training, accent correction, etc. generally replay sample audio data recorded on magnetic tape, etc., and the vocal trainer opens the audio data and speaks. A method of training has been implemented.

A step forward in these training methods is to compare and display the vocal patterns of the 2nd vocal trainee and the vocal pattern of the model, and then change the vocalization method so that it becomes a vocal pattern that is very similar to the vocal pattern of the 1st model. There is a way to train them.

In either case, the evaluation and judgment of the effectiveness of the training is made by the vocal trainee himself, so if the vocal trainee's own judgment is poor or incorrect, the training may be performed without the correct vocalization. Put it away. Therefore, it has been desired to put into practical use a vocal training device that can objectively determine whether the vocalization method is correct and stable, and can perform vocal training efficiently.

[Conventional technology]

As a conventional technique, a vocal training device having a function of comparing and displaying the vocal pattern of two vocal trainees and the vocal pattern of a model will be explained as an example.

FIG. 2 shows a conventional vocal training device having a display function.

In the figure, a is a label for classifying a vocal unit, b is a pinch pattern of a vocal unit, C is voice data of a vocal unit, and d is a model voice pattern.

e indicates each trainee's voice pattern.

Also, Figure 2.

1. Keyboard for specifying training vocal units, changing patterns, etc. 2. Microphone for inputting voice trainee's vocalizations 2. Audio player for reproducing audio 3. Control unit for controlling the operation of the entire device 4. Audio data A feature extraction unit 5 for extracting features (pinch patterns); a voice pattern storage unit 6 for storing a model voice pattern; a display device 7 for comparing and displaying the model voice pattern and the voice pattern of the voice trainer. It is composed of ゜.

The acoustic information handled in this explanation is a pitch pattern, and the audio pattern storage unit 6 stores audio data C of all kinds of model vocalizations and their pitch patterns b. It is also assumed that the display section of the display device 7 is divided into a model voice pattern display section d and a voice training person's voice pattern display section e.

When practicing standard pronunciation using this voice training device, the voice pattern (pinch pattern b) of the word to be trained to voice (in this example, "Tokyo") is retrieved from the voice pattern storage unit 6 by specifying on the keyboard 1. 1 is displayed on the model voice pattern display section d of the display device 7.

Next, the vocal trainer imitates the model vocalization and repeats "Tokyo" in the same way.
5 This voice is analyzed and the pitch pattern of the voice trainee's voice is displayed in the voice pattern display section e of the voice trainee.
are displayed in parallel. Vocal trainers compare both pitch patterns and observe the differences.

When he decides that there is a difference, he says ``Tokyo'' again, and
Compare it with the new pitch pattern.

[Problem that the invention seeks to solve]

As mentioned above, conventional methods only display the pitch pattern that is analyzed each time you speak, so once a well-crafted pitch pattern appears, you can check whether you have really mastered the vocalization method by trying it out repeatedly. The voice trainees made their own decisions when they were satisfied.

However, as mentioned above, since it is not an objective judgment, there is a problem in that it is not clear whether or not the student has really become able to produce standard pronunciations stably after using the seven model pronunciations.

[Means for solving problems]

The present invention aims to realize a new vocal training device that solves the above-mentioned problems.

The problem is that there is a storage means for storing a plurality of vocalization patterns of one voice trainee, a statistical processing means for statistically processing the vocalization patterns stored in the storage means, and a method based on statistical values by the statistical processing means. The problem is solved by the voice training device according to the present invention, which is provided with a determining means for determining and evaluating the stability of the voice of the voice trainee.

[Effect]

In other words, a certain dark value is set to objectively judge the results of vocal training, and the difference (distance) is determined by comparing the multiple vocal pinch patterns of one vocal trainee with the latest vocal pitch pattern of the vocal trainee. ).

Find the variance of these distances.

When this variance value falls below the set threshold.

It is judged and displayed that stable results have been achieved for the first time.

〔Example〕

The gist of the present invention will be specifically explained below with reference to an embodiment shown in FIG.

FIG. 1 shows a block diagram showing one embodiment of a vocal training device according to the present invention. In the figure, f is the dark value display portion and 3g is the dispersion value display portion.

h indicates a determination result display portion, respectively.

The same symbols indicate the same objects or contents throughout the figures.

This embodiment uses the conventional components explained in FIG.

A dark value storage section 8 defines the limit of the variance value that can be used to determine that the vocalization is stable.A reading section 9 determines whether the variance value is within the dark value.
゜Distance calculation section 10゜Calculates the distance between pitch patterns of voice data ゜Dispersion calculation section 11゜Calculates the dispersion state of a plurality of old calculation values calculated by the distance calculation section 10゜Dispersion calculation section 11゜Distribution calculation section 11゜Calculates the dispersion state of a plurality of old calculation values calculated by the distance calculation section 10゜Distribution calculation section 11゜Distribution calculation section 11゜Calculates the dispersion state of a plurality of old calculation values calculated by the distance calculation section 10 It consists of a pitch pattern vanofa 12° for storing pitch patterns.

In this embodiment, the display device 7 displays a pinch pattern of the model voice data and a picture of the voice data of the vocal trainee.

The distribution pattern, dark value 1 dispersion degree 9 judgment results, etc. shall be displayed. Also, in this embodiment, it is assumed that the user is trained to pronounce "Tokyo".

By the specified manual input on the keyboard 1, the pi-nochi pattern b based on the model voice data C of "Tokyo" is displayed on the display section W7, and at the same time, the model voice data C is outputted from the voice reproduction section 3. The vocal trainer imitates the vocalization of model voice data C and says,
When you say "Tokyo".

The voice data is analyzed by the feature extraction section 5, and the two analysis results are displayed side by side with the pitch pattern b of the model voice data C.

The voice trainer compares both patterns, observes the difference, and utters "Tokyo" again, and the voice trainer's pattern on the display device 7 is replaced with a pattern obtained by analyzing the new voice data.

The voice trainer repeatedly uttered ``Tokyo.''

I devised my vocalization method so that my own pattern resembled the pattern of the model audio data, and then used pattern b of the model audio data C.
Once you have determined that it is similar to , repeat it several times to confirm the way you pronounce it at that time.

Since the pitch pattern bannohua 12 stores the latest pattern repeatedly uttered in the above state and the pitch patterns of several times before that, the latest pattern can be retrieved from the data stored in the pitch pattern bannohua 12. The distance calculation unit 10 calculates the distance between the pattern and the previous several patterns, and the variance calculation unit 11 calculates the variance of these distances.

When this variance value falls below a dark value that is preset and stored in the dark value storage unit 8, the determining unit 9 determines that stable vocalization has started. This determination is made every time a new utterance is made, and the threshold is 1 variance.

The determination results are displayed in sections f, g, and h on the display device 7, and are updated each time a new utterance is made.

When a vocal trainer is searching for a vocal method, the vocal pattern changes greatly each time, so the variance value becomes large.

It is judged to be unstable, but if you find a vocalization method and start practicing repeatedly, the 1 variance value will become smaller and eventually it will be judged as stable. If stable is displayed, you will have mastered the vocalization method for "Tokyo". Let's move on to the 5th order voice training.

Note that the threshold value can be changed according to the wishes of the voice trainer.

Next, other embodiments will be described.

FIG. 3 shows another embodiment of the vocal training device according to the present invention. In the figure, e' is the voice pattern of the trainer's utterance that is judged to be correct by the instructor, r' is the average darkness value display part, f'' is the distributed darkness value display part, and j is the average value display part. are shown respectively.

The embodiment shown in Figure 1, whose number is not shown, is intended for use by a vocal trainer who is self-studying from the beginning.

This embodiment is based on the assumption that the instructor once judges the utterances of the voice trainee and then the voice trainee studies on his or her own.

This embodiment is of the constituent parts shown in FIG.

A variance calculation unit 12 calculates the average value and variance value of the values calculated by the distance calculation unit 10.An average/variance calculation unit 13°.A dark value storage unit 8 stores the dark value of the average value and the dark value of the variance value. The only difference is that the average/dispersion value storage section 14° has been changed, and the other parts have the same configuration.

The contents stored in the voice pattern storage section 6 also include the voice data patterns of the voice trainee who are determined to be correct by the instructor from the model voice data patterns.

Also, the display contents on the display device 7 are in addition to the darkness value 9 dispersion value that is displayed in the embodiment of FIG.

The dark value of the average value, the dark value of 1 variance value, the average value 2 variance value, f',
It is now displayed in the f″ and j parts respectively.

This example calculates the distance between the voice data pattern determined to be correct by the instructor and the vocalization patterns uttered several times by one voice trainee during self-study, calculates the average value and variance value, and calculates the darkness value of each. When the value is lower than the darkness value, it is determined that the voice data pattern determined to be correct can now be uttered stably.

〔Effect of the invention〕

According to the present invention as described above, a high training effect can be obtained by objectively evaluating the vocal training effect of a vocal trainee using the stability of vocalization as a measure.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of a vocal training device according to the present invention. FIG. 2 shows a conventional vocal training device with a display function. FIG. 3 shows other embodiments of the vocal training device according to the present invention. In fig. 1 is the keyboard, 2 is the microphone. 3 is an audio playback section, and 4 is a control section. 5 is a feature extraction section; 6 is a voice pattern storage section; 7 is a display device; and 8 is a dark value storage section. 9 is a determination section, and 10 is a distance calculation section. 11 is a distributed calculation section. ]2 is a pitch pattern buffer section. 13 is an average/variance calculation unit 1; 14 is an average/variance value storage unit; are shown respectively. Blessed with drunkenness 2

Claims (1)

[Claims] In an apparatus comprising a function of analyzing uttered voice data, extracting acoustic information and creating a vocalization pattern, and a function of displaying the vocalization pattern extracted and created by the function, a plurality of voice trainers storage means for storing the vocalization patterns of the vocal trainee; statistical processing means for statistically processing the vocalization patterns stored in the storage means; and stabilization of the vocalization of the vocal trainee based on the statistical values by the statistical processing means. 1. A vocal training device comprising a determination means for determining and evaluating gender.