JPS6177100A - Voice section detecting circuit - 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 The present invention relates to a speech interval detection circuit in a speech recognition device.

In general, when a speech recognition device uses speech with a good signal-to-noise ratio as an input speech, it is relatively easy to extract a section where speech exists. However, the environment in which the voice recognition device is actually used includes various kinds of noise, and the voice is input in a form superimposed on the noise. Since the noise at this time changes from moment to moment, it is difficult to detect a stable voice section by setting a fixed threshold value and cutting out the voice section, which may cause misrecognition. When clipping is performed using such a fixed threshold value, under high noise conditions, noise is added before and after what should normally be a voice section and is clipped.

1-1 The present invention has been made in order to solve the above-mentioned problems, and in particular, to perform stable speech segment detection regardless of the level of ambient steady noise to ensure a stable recognition rate. It was done for that purpose.

Request-a In order to achieve the above object, the present invention provides a voice section detection device that extracts voice signal power and sets a threshold value within the silent section to cut out the voice section.

Sample hold that samples and holds the noise level at that time using a pulse that occurs when a predetermined time (t seconds) has elapsed from the end of the voice and a pulse that occurs after a predetermined time (S seconds) has elapsed from the start of the voice. The present invention is characterized in that it has a means for detecting a voice section and uses the hold value as a threshold for cutting out a voice section. Hereinafter, the present invention will be explained based on examples.

The present invention holds the noise level at the time when a predetermined time (t seconds) has elapsed from the end of the voice section and the noise level at the time when a predetermined time (S seconds) has elapsed from the start of the voice section, and converts the noise level into a voice. This is used as a threshold for section detection. The above t is 200 silent intervals for words with consonants.
This is based on the fact that ~400+ns exists, and the decision is made with the consideration that if the next voice is inputted after this time, the voice that is around 1 will be processed as one word. Also, if the level of environmental noise increases during vocalization, it may not be possible to detect the end, so if the end cannot be detected even after S seconds have passed from the start, that point will be forced to be the end of the voice section. and hold the noise level at that time.

FIG. 1 is an electrical block diagram for explaining an embodiment of the voice section detection circuit according to the present invention, and FIG. 2 is a time chart. In FIG. 1, 1 is an input terminal, and 2 is a detection circuit. , 3 is a smoothing circuit, 4 is a sample hold circuit, 5 is a level comparison circuit, 6 is a voice section discrimination circuit, and 7 is an output terminal. Further, in FIG. 2, (Q) shows an example of the average signal level of the input signal, and T1. T2 indicates the switching point of the threshold value. (b) is the voice section signal detected using the threshold of (0), (c) is the voice section end pulse that occurs when the section signal is low t seconds after the fall of the voice section signal in (b), (d) is a pulse that occurs when the terminal pulse in (c) does not occur within S seconds from the rise of the voice section signal in (b), and the sample hold moves due to the pulses in (C) and (d), and the threshold value is set. Switch.

In FIG. 1, the average signal level of an input signal from an input terminal 1 is detected through a detection circuit 2 and a smoothing circuit 3, and is input to a sample hold circuit 4 and a level comparison circuit 5. The output signal from this level comparison circuit 5 is input to a voice section discrimination circuit 6.

If the voice section end pulse generated here and the end cannot be detected, the pulse generated S seconds after the start end is input to the sample hold circuit 4, and the signal level at that time is held. This held value and the average signal level are compared by one in the level comparison circuit 5 and outputted to the voice section signal stage 7. Note that the above method uses the noise level at the time when the audio end pulse and the pulse that occurs when the end cannot be detected as the next threshold, but in order to operate more stably, a fixed value for this threshold can be set. A method may also be considered in which the added value is used as a threshold for voice section detection.

FIG. 3 is an electrical block diagram showing an embodiment of the above method. In this embodiment, the threshold value for voice section detection is the sum of the above threshold value and a fixed value. 1 is a reference voltage source, 9 is an adder, and other parts having the same functions as in FIG. 1 are given the same reference numerals as in FIG. Thus, this embodiment has a reference voltage source 8 and an adder 9, and the adder 9 adds a constant bell voltage Vs from the reference voltage source 8 to the output of the sample and hold circuit 4.
This added value becomes the new threshold value of the level comparison circuit 5.

As is clear from the above description, according to the present invention, it is possible to detect an interval signal by determining an optimal threshold value.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is an electrical block diagram for explaining an embodiment of the present invention, and FIG. 2 is a time chart. FIG. 3 is an electrical block diagram showing another embodiment of the present invention. 1... Input end, 2... Detection circuit, 3... Smoothing circuit, 4... Sample hold circuit, 5... Level comparison circuit, 6... Voice section discrimination circuit, 7... Output Edge, 8・
...Reference voltage, 9...Adder. If bad Figure 2 Figure 3

Claims (2)

[Claims] (1) In a voice section detection device that extracts the voice signal power and sets a threshold value within the silent section to cut out the voice section, a pulse and a It has a sample hold means that samples and holds the noise level at that time using a pulse generated at the time when a predetermined time (s seconds) has elapsed from the start of the voice, and uses the hold value as a threshold for cutting out the voice section to separate the voice section. A voice section detection circuit characterized by detecting. (2), comprising means for adding a fixed value to the threshold;
The speech section detection circuit according to claim 1, wherein the speech section is detected using the value as a threshold for cutting out the speech section.