JPS6078471A - 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 speech training device that extracts feature signals of speech accompanying speech and visualizes the feature signals on a display device for speech training. .

Configuration of the conventional example and its problems Conventionally, in a speech training device that performs speech training by visualizing the characteristic signal -υ of each voice on the display screen of the display device, from the display start point to the display screen of the display device, One frame is constructed by allocating a predetermined 11.11 period (for example, 5 seconds) necessary for displaying one spoken word, and the extracted voice characteristic signal is Jl:
U! Displays sequentially from the display start end point to the end point of the frame,
After reaching the end point, 11 and the display starts if'i! 17 < is displayed at the ji point at a predetermined time 11 of 1 arene, but the 4 and v signals are displayed at a fixed position of 1 from the display start point to the end of 1 frame. However, in order for the speaker to time the utterance himself, for example,
People with disabilities and people with disabilities may speak too quickly, or if they hear one utterance within a given frame. There may be cases where the payment is not completed.

Furthermore, since the display is repeated cyclically, I was nervous when trying to figure out the timing, and when I made a mistake in the timing of my speech, I had to get to the display start point of the next frame, which put pressure on my speech. There is a problem of giving.

OBJECTS OF THE INVENTION The present invention solves the above-mentioned drawbacks of the conventional art.It is an object of the present invention to provide automatic discrimination means for speech and non-speech based on characteristic signals of speech, and to detect the "-" characteristic by the start of speech by the speaker. To provide a speech training device in which one frame of a signal is displayed so that speech timing can be easily adjusted and speech information can be easily stored within 17V.

Structure of the Invention The present invention provides a feature extracting means for extracting a characteristic signal of a speaker's voice, a storage means for storing the characteristic signal of the voice, and an utterance discrimination device for determining the start of speech from the characteristic signal of the voice. means, dividing a display range of one frame on the time axis in the non-speech state output of the utterance determining means, and shifting and displaying the voice characteristic data in one section of display range divided from a display start point; The speech training device is equipped with a display control means for displaying one frame on the time axis extracted from the display range in the speech state output of the speech discriminating means, and the speech training device is equipped with a display control means for displaying one frame on the time axis by extracting the speech state output from the speech discriminating means, Display the previous feature signal,
By continuously displaying the characteristic signal dynamically toward the terminal point by the utterance a7i of a regular speaker, the utterance i, i
11. It is possible to easily match the display timing and display the utterance information within a predetermined display solenoid.
Regarding the rising edge information of the feature signal before speech, also on Tuesday 11.11
It is possible to maintain and display 1 bit in between.

DESCRIPTION OF EMBODIMENTS FIG. 1 shows a speech training device 1 according to an embodiment of the present invention.
1 is a block diagram showing the configuration of IJ. 1 is a microphone 1a, a vocal fold motion sensor 1b, a valve vibration sensor 1C1 an exhalation flow sensor 1dj; This is a feature extraction circuit that extracts a feature signal. 2I''i storage means for storing data for one display frame of the characteristic signal; 3 utterance discrimination means for determining the start of the speaker's utterance from the q'!f characteristic signal of the feature extraction circuit 1; , 4 represents the feature signal in X-Y coordinates 4;
9. A display device such as a television for displaying graphic images; 5 is a feature signal of the feature extraction circuit 1; and utterance discrimination means 3.
The display control means receives the output signal i as an input, and stores voice characteristic data in the storage means 2 in correspondence with the input signal i and controls the display of the data on the display device 4.

FIG. 2 shows an example of how the display screen of the display device 4 is drawn.
In the range from the display start end point 6 to the display end point 7 in the horizontal direction, the time required for one shot HQ to be manually set is set, and the display resolution specified by the required sunblink period of the viewing feature extraction signal is set. Display the characteristic signal. FIG. 3 shows a diagram of the output waveform of the feature extraction circuit 1. Here, all 4! of this signal are shown. 'r is the human input signal of the discriminating means 3. FIG. 4 shows the storage format of the storage means 2.

This embodiment is structured as follows. The operation of L1 training equipment will be explained below.

The display control means 6 inputs the feature signal which is the output of the feature extraction circuit 1 at a predetermined sampling period and stores it in the storage means 2 as a fourth signal.
When data is stored sequentially from the storage area corresponding to the display start end point 6 in the format shown in the figure, data is sequentially displayed from the display start end point 6 on the display screen. When the characteristic value after rw is not determined to be a speech state by the speech discrimination means 3 (more than the voltage Vth in FIG. 3), the display control means 5 displays the value 1; data (Fig. 4) in the direction of the starting end point.
The data is shifted and the T on the display screen of the display device 4 is
The display of w's display area is lowered. The second process is repeated during the period in which the output of the speech discrimination means 3 is detected to be in a non-speech state.Here, as shown in FIG. When the output of is detected as a speech state, the display control means 6d extracts the table screen j'' of the section of Tw and executes the display process in the direction of the terminal point 70 of the display screen. .

FIG. 6 is a flowchart showing the following operations. Each step □ will be explained below.

(b): This is the initialization step, and the input /l/
Set the number n=o.

(b): When the feature data D++ is input manually, this data is stored at address An of the storage means 2.

First data D. is address A. is stored in

(C): Determine whether the time -1n from the start of the operation corresponds to the divided display period 'rw of the screen of the display device 4.

): Address Ay when tn is not tw
The data Dn stored in is displayed at the desired screen position by 4 degrees on the time axis tn. Address people. Data D stored in . is time t. That is, it is displayed at display start position 6.

(e): Move on to next sample processing.

Thereafter, (b) to C→ are repeated until tn≧tw.

(to): Data Dn is a predetermined value Dth (in FIG. 3, Vu+
). That is, fa l;
Determine whether A has been performed.

(G): Dn (In case of Dth, that is, when no utterance has been made, shift the data stored at address A.-Aw by one), move it to o -Aw-, (as a result, to A. (The stored data will be discarded.)

(b) Near address A. -Display the data stored in AV-1 (displayed during the period Tw shown in Figure 2)
. Return to 1., (b).

(S): When Dn≧Dth, that is, when a word is uttered, data Dn is displayed at position 11.1i1t.

): Move to next sample processing.

(/→: Determine whether or not the time tn corresponds to 4.11 in one frame.

(P): When 1n≦tk, stores the next data Dn,
Return to (su).

Therefore, in -1-lQl, 1, until tn>tk (
(S) ~ (W) is the edge Jly, and is drawn.

How can you determine the speech state in Figure 1, if, 7i'Fl group 1, stage 3?゛↑Voice intensity or throat vibration envelope: If done by ascending,
The discrimination of all utterances is IINjL. Additionally, the rise of other 1.1 characteristic parameters such as the nasal vibration envelope may be used.

As described above, according to this embodiment, by providing the utterance determination means 3, the utterance timing of the speaker is automatically displayed at a predetermined position on the display screen, thereby eliminating unnecessary intervention by the speaker. do not need. This is also effective for comparison training with a model display of speech. In addition, low-level signals that are not detected as speech are stored and displayed on the display screen and part of the memory area, so it is possible to display the rise characteristic from a low level before speech starts in real time. .

Effects of the Invention The utterance 5-1 training device of the present invention achieves real-time performance by providing means for automatically determining the start of speech and display control means for displaying pre-utterance data in some divided areas. Even though I have
Since data from the state before and after speech to the end of speech can be easily set within one frame on the display screen, it is not necessary for the speaker to time the speech, and this has a great practical effect.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of a speech training device according to an embodiment of the present invention, Fig. 2 is a display screen of the same embodiment...1 format diagram, Fig. 3 is a waveform diagram showing an example of a characteristic signal, and Fig. 4 6 is a format diagram showing the storage former of the storage means, 1.
The figure is a flowchart showing the operation of the embodiment described above. 1...! 1.+J oil extraction circuit 2...Storage means, 3...Utterance discrimination means, 4...Table 7'1
．． : device, 5...display control means. 14 River Ill Yube Figure 2 Figure 3 Figure M4

Claims (1)

[Claims] a feature extracting means for extracting a characteristic signal of a speaker's voice; a storage means for storing the characteristic signal of the voice; an utterance determining means for determining the start of speech from the characteristic signal of the voice; In the output of the non-speech state of the means, the display range of one frame on the time axis is divided, and the characteristic data of the voice is shifted 1 in the display range of a certain section divided from the display start point.
-Table, :; 17. A ttf training characterized by comprising display control means for displaying one frame on the time axis by extracting from the display range in the speech state output of the speech discriminating means. Device.