JPS6242200A - Voice recognition equipment - 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] technical summer disaster The present invention relates to a speech recognition device.

It is widely known that in conventional speech recognition devices that recognize multiple different types of voices in one place, all types of voices take the same utterance interval, but this makes it difficult to classify different types of voices, and it is difficult to classify different types of voices. You can't get a rate.

In addition, the technology that utilizes vocalization intervals is disclosed in Japanese Patent Application Laid-Open No.
Although it is known in the speech input document creation device disclosed in Publication No. 04L, the utterance interval in this case is used to separate words or sentences, and is not used to classify different types of speech. Furthermore, if different types of speech, such as words and monosyllables, are recognized using the same method, the recognition accuracy will deteriorate, so it is common to have separate recognition units even though the input unit is the same. Generally, the length of the input audio.

It is classified based on spectral changes in the power stable part.1 For example, the single syllable "re" and the word "example (re-)" and the sound collecting part "re-" are classified.
It has the disadvantage that it is not possible to accurately distinguish between words such as "kiyo" and American "kyou".

(2) The present invention was made in view of the actual situation such as "Double Recognition", and in particular, it was made for the purpose of providing a speech recognition device with a high recognition rate.

In order to achieve the above object, the present invention provides a speech recognition device that includes a speech input section and a speech recognition section and that recognizes a plurality of different types of speech. When the type of audio differs, the utterance interval is different from the utterance interval that is uttered at the same time as the type of the previous input audio, or the sound collection unit detects the part related to the audio from the collected signal. In a speech recognition device comprising a detection unit capable of recognizing a plurality of different voices, and a recognition unit capable of recognizing a plurality of different voices, the time interval between the voice to be recognized and the voice recognized before and after (is measured, and the time interval is The recognition section of the audio signal is selected depending on the length of the sound signal, or the sound collection section and
A speech recognition device comprising a detection unit that detects a voice-related part from a collected signal, a recognition unit that can recognize different types of voice, and a display unit that displays recognition results. The present invention is characterized in that the time interval between the voice recognized before and after the voice recognized is measured, and one of the recognition results is selected and displayed depending on the length of the time interval. Hereinafter, the present invention will be explained based on examples.

The present invention provides a speech recognition device that recognizes a plurality of different types of speech, and when the type of speech to be input is different from the type of the immediately preceding input speech, the interval between utterances when the type of speech to be input is the same as that of the immediately preceding input speech is provided. By making the utterance interval longer than the utterance interval, it is made easier to classify the type of voice, thereby increasing the recognition rate.

FIG. 1 is a block diagram for explaining one embodiment of the present invention. In the figure, ] is a voice input section, 2 is a voice recognition section, 3 is a Hataki section, and the voice recognition section 2 is a word recognition section. 2A and monosyllable recognition section 2B
The voice input section 1 inputs monosyllables and words.

FIG. 2 is a diagram showing the overall processing flow in the speech recognition device, where tl is the utterance interval when single syllables or words are input continuously, and tl is the utterance interval when monosyllables and words are input alternately. However, t2.

If you make sure that the beginning of the sentence you input is a monosyllable, for example, "tenki" in the sentence "Kiyouba Tenki ga goodesu".

When the three words "good" and "desu" are registered in the word dictionary pattern, the utterance is tt tt t2 t2 t2 tt
You can do it like this.

FIG. 3 is a diagram showing a modified embodiment of the embodiment shown in FIG. 1. In the embodiment shown in FIG. is provided with a specifying section 4, which specifies whether the first input speech is a monosyllable or a word.

4 and 5 are block diagrams for explaining other embodiments of the present invention, respectively. In the figures, 11 is a sound collection section, 12 is a detection section, and 13 is a time interval measurement section.

14 is a word recognition unit, 15 is a monosyllable recognition unit, and 16 is an output display unit. In this case, for example, when inputting a monosyllable, input the interval from the previous utterance to jl + word. time, let the interval between the previous utterance be t2 (
However, t1≠t2).

Now, if we input the example sentence of the example shown in Figure 1 under the same conditions (it contains three words), the way to pronounce it at that time is tt t, t2 t, t2 t2
(In this example, tl<t2).

In the embodiment shown in FIG. 4, the time interval between the speech to be recognized and the speech recognized before and after the speech is measured by the time interval measuring section 13, and the length of the time interval is used to determine the sound signal. Selection of recognition units, namely word recognition unit 14 and monosyllable recognition unit 1
In the embodiment shown in FIG. 5, one of the recognition results is selected depending on the length of the measured time interval.

Effects As is clear from the above explanation, according to the present invention, in a device that recognizes a plurality of different types of voices, it is possible to classify the types of voices without providing a special circuit for classifying them. A high-performance recognition device can be obtained at low cost.

[Brief explanation of the drawing]

FIG. 1 is a block diagram for explaining an embodiment of the speech recognition device according to the present invention, FIG. 2 is a flowchart for explaining the operation of the embodiment shown in FIG. 1, and FIG. 1st
Figures 4 and 5 showing modified embodiments of the embodiment shown in the figures are configuration diagrams showing other embodiments of the present invention, respectively. DESCRIPTION OF SYMBOLS 1... Voice input part, 2... Recognition part, 2A... Word recognition part, 2B... Monosyllable recognition part, 3... Output part, 4
... designation section, 11 ... sound collection section, 12 ... detection section,
13... Time interval measurement unit, 14... Word recognition unit, 1
5... Monosyllable recognition section, 16... Output display section. Patent applicant: Ricoh Co., Ltd. No. 1 Figure 2 Figure 3 Figure 4 Figure 5 Procedure amendment band (master) 1985 Patent application No. 182521 2 Name of invention Speech recognition device 3 Case of person making amendment Relationship with Patent applicant Ri Ota Nakamagome Address 1-3-6 Nakamagome, Ota-ku, Tokyo Name (674) Ricoh Co., Ltd. Representative Hama 1)
Hiro 4, Agent Address: 1-2-7 Furocho, Naka-ku, Yokohama 231
6. Subject of amendment Column 7 for detailed explanation of the invention in the specification
, "By lengthening the utterance interval" stated in the first line of page 5 of the Specification of Contents of the Amendment is amended to "by lengthening (or shortening) the utterance interval."

Claims (4)

[Claims] (1) In a speech recognition device that has a speech input section and a speech recognition section and recognizes a plurality of different types of speech, when the type of speech to be input is different from the type of the previous input speech, the A speech recognition device characterized in that the speech recognition interval is different from the speech interval that is produced at the same time as the type of input speech. (2) The speech recognition device according to claim (1), characterized in that the type of first input speech is specified. (3) In a speech recognition device comprising a sound collection section, a detection section that detects a voice-related part from the collected signal, and a recognition section that can recognize a plurality of different voices, A speech recognition device characterized in that a time interval between a speech recognized before and after the speech recognized is measured, and a speech signal recognition unit is selected based on the length of the time interval. (4) A sound collection unit, a detection unit that detects a voice-related part from the collected signal, a recognition unit that can recognize different types of voice, and a display unit that displays recognition results. In a speech recognition device consisting of a speech recognition device, the time interval between the speech to be recognized and the speech recognized before and after it is measured, and one of the recognition results is selected and displayed depending on the length of the time interval. Characteristic voice recognition device.