JPH1185456A - Conversation recorder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to simply and accurately correct errors in voice recognition even after a certain time has passed from the end of a conversation or a conference, in a conversation recorder utilizing a voice recognition technology. SOLUTION: The conversation recorder is provided with a voice input means 1, a voice recording means 2 for recording voice, a character string conversion means 3 for recognizing voice and converting the voice into a character string, a character string recording means 5 for recording character strings, and a connection recording means 6 for recording corresponding recording addresses of the means 2, 5 and constituted so as to listen the original voice of the voice- recognized character string. Thereby the conversation recorder can easily and accurately correct recognition errors even when memory becomes nuclear after a certain time has passed time after the end of a conversation.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a conversation recording apparatus for recording the contents of a conversation or a conference in the form of characters.

[0002]

2. Description of the Related Art Hitherto, several methods have been used for recording conversations. The easiest way is to create a record based on the memories of the participants in the conversation. In addition, a person who participates in the conversation while conducting the conversation, or a dedicated person who records the conversation, often uses a method such as a memo. In particular, the latter is a method performed at important meetings, such as the Diet.

As a method of recording a conversation, there is a method of using a recorder or a video tape recorder. JP 8
Japanese Patent Application Publication No. 194492 discloses a method combining voice recording, text formation means, display means and the like. The writing unit converts the contents of the meeting into a character string by using a keyboard or voice recognition and records it.

[0004]

When recording a conversation or a meeting, the method of documenting the document is excellent in recordability and preservation. However, when a record is created by relying on the memory of a participant, there may be cases where a record covering the contents of the conversation cannot be created due to forgetting or the interest of the person who creates the record. Taking notes can reduce forgetfulness, but the content may be lost depending on the interest of the recorder. These problems can be eliminated if a full-time recorder is used, but it is not easy because a dedicated person must be prepared.

The method disclosed in Japanese Patent Application Laid-Open No. 8-194492 may be effective in the case where an effective progress of a conference or a strict record of a conference is to be recorded, but it is not easy to record a conversation. When speech recognition is used, the text of the record is automatically written. However, since the recognition is incorrect, it is necessary to correct the recognition error. If this correction operation cannot be easily and correctly performed, it takes a long time to create a record.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a conversation recording apparatus using a speech recognition technique, in which a speech or a conference can be corrected easily and accurately even after the end of a conversation or a meeting and the memory is lost. I do.

[0007]

SUMMARY OF THE INVENTION In order to solve this problem, a conversation recording apparatus according to the present invention comprises a voice input means, a voice recording means for recording voice, and a character string for recognizing voice and converting it to a character string. Conversion means, sentence separation detection means for detecting a break in the converted character string, character string recording means for recording the converted character, and concatenated recording for recording the corresponding recording addresses of the voice recording means and the character string recording means Means, a sound reproducing means, a character string display means, a character string position indicating means, and a character string correcting means for correcting an erroneous character string.

[0008] This makes it possible to record the conversation or the conference easily and accurately even after the conversation or the conference ends and the memory is lost after a lapse of time.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of a conversation recording apparatus according to the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a block diagram of a conversation recording apparatus according to an embodiment of the present invention. In FIG.
2 is a voice recording means, 3 is a character string conversion means, 4 is a sentence separation detecting means, 5 is a character string recording means, 6 is a connection recording means,
Reference numeral 7 denotes voice reproducing means, 8 denotes character string display means, 9 denotes character string position indicating means, and 10 denotes character string correction means.

The operation of the present embodiment will be described separately with reference to the drawings at the time of recording a conversation and at the time of correcting a character string which has been recognized incorrectly.

First, the operation during conversation recording for recording a conversation will be described. The voice of the conversation is converted by the voice input unit 1 into an electric signal that can be handled by other components. The output of the voice input means 1 is applied to the voice recording means 2 and the character string conversion means 3. The audio recording means 2 sequentially stores the added audio signals. The character string converting means 3 recognizes the voice signal by voice and converts it into character string information. The converted character string is provided to the character string recording means 5 and the sentence separation detecting means 4.

The sentence break detecting means 4 detects the break of the sentence of the conversation based on the words and the like in the character string output by the character string converting means 3 and outputs the leading time of the phrase as a detection result. The head time of the phrase is added to the link recording means 6. The concatenated recording means 6 records the character string obtained by converting the same phrase as the head time of the phrase into the character string recording means 5 together with the storage position. Thereby, the contents of the voice recording means 2 and the contents of the character string recording means 5 can be related. That is, by specifying a character string whose conversation content has been converted, the position of the corresponding sound can be known, and the sound can be reproduced.

Next, the operation at the time of correcting a character string having an incorrect recognition will be described. The character strings stored in the character string recording unit 5, that is, the contents of the conversation converted into the character strings, are stored in the character string display unit 8.
Is displayed with. By checking the displayed contents, it is possible to know the location where the recognition error has occurred. The character recognition position is specified by the character string position designating means. The designated position of the character string is added to the connection recording means 6 and the character string correction means 10. The link recording means gives the voice recording means 2 the position of the voice recording means 2 corresponding to the designated character string. At the same time, the contents stored in the sound recording means 2 are output as sounds from the sound output means 7. While watching the character string on the screen, the operator simultaneously listens to the sound output from the sound reproducing means 7 and operates the character string correcting means to correct the character string in the character string recording means 5. As described above, the character string can be corrected using the character string correction means while listening to the character string and the corresponding sound, and there is no need to guess the correct character string from the character string converted due to the recognition error. Can be kept accurate.

Although the output of the voice input means 1 is added to the character string conversion means 3, the voice recording means 2 may be controlled and used as input for voice recognition. With this configuration, the voice stored in the memory can be processed, so that when the conversion time of the voice recognition is slow, the voice recognition can be delayed and not performed in real time.

As a method for realizing the voice compression means 2, a voice compression technique can be applied. FIG. 2 shows a block diagram in which the audio compression technique is applied to the audio recording means 2. 2, reference numeral 2 denotes the audio recording means shown in FIG. 1, 11 denotes an audio signal compression means, 12 denotes a compressed audio recording means, 13 denotes a compressed audio expansion means, 21 denotes an audio input signal, 22 denotes an audio reproduction output signal, 23 is a voice recording means control signal. The voice recording means of FIG. 1 is shown again in the upper part of FIG. The lower side shows a block diagram to which the audio compression technique is applied. Replacing the upper part with the lower one is an embodiment of the present invention in which audio compression is adapted. When the audio compression technique is applied, the size of a memory required for recording is reduced, and the present invention can be used for a device that has severe restrictions on cost, size, and the like.

Generally, after the recognition error is corrected, the contents of the voice recording means 2 and the connection recording means 6 become unnecessary.
If only the capacity of the character string storage means 5 is increased, many records can be held. Apart from such usage, it is also possible to leave only necessary parts of the voice recording means 2 and the connection recording means 6 and delete other parts. By doing so, it is possible to save only the voice of the part of the conversation that the user wants to keep together with the document.

As described above, according to this embodiment, a recognition error can be easily corrected even after the conversation has ended and the memory has become ambiguous after a lapse of time.

[0019]

As described above, according to the present invention, voice input means, voice recording means for recording voice, character string conversion means for recognizing voice and converting it to a character string, and converted character string Sentence delimiter detecting means for detecting a delimiter, character string recording means for recording the converted character, voice recording means and concatenated recording means for recording the corresponding recording address of the character string recording means, voice reproducing means, By composing the string display means, the character string position designating means, and the character string correcting means for correcting an incorrect character string, a simple recording of the conversation can be performed, and the storage is ambiguous after a lapse of time after the conversation is over. Even after that, the recognition error can be easily and accurately corrected, which is very convenient in practice.

[Brief description of the drawings]

FIG. 1 is a block diagram of a conversation recording device according to an embodiment of the present invention.

FIG. 2 is a block diagram of a sound recording unit to which a sound compression technique is applied;

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 voice input means 2 voice recording means 3 character string conversion means 4 sentence break detection means 5 character string recording means 6 connection recording means 7 voice reproduction means 8 character string display means 9 character string position indicating means 10 character string correction means 11 voice signal Compression means 12 compressed sound recording means 13 compressed sound expansion means 21 sound input signal 22 sound reproduction output signal 23 sound recording means control signal

Claims (1)

[Claims] A voice input unit for inputting a voice of a conversation; a voice recording unit for recording an output of the voice input unit; a character string converting unit for recognizing a voice of the output of the voice input unit and converting the output into a character string. A sentence separation detecting unit that receives an output of the character string conversion unit and detects a break of the converted character string, a character string recording unit that records an output of the character string conversion unit, and an output of the sentence separation detection unit Receiving a corresponding recording address of the voice recording means and a corresponding recording address of the character string recording means, a coupling recording means, and a voice reproducing means capable of reproducing the recorded voice of the voice recording means from a designated position, Character string display means for displaying a character string recorded in the character string recording means; character string position designating means for specifying the position of the character string to be displayed by the character string display means; A conversation recording device comprising a character string correcting means for correcting a column.