EP0175503A1
Expired / Lapsed
EP
⏱️ 0.0 years remaining
Method and apparatus for use in interactive dialogue
Assignee: British Telecommunications PLC
•
Filing: 1985-08-23
•
Granted: —
•
Expires: 2005-08-23
Abstract
Abstract A dialogue controller (15) and a speech recogniser (11) are provided for obtaining classified information from utterances spoken by a user in a dialogue. Templates to enable the recogniser to recognise utterances are grouped in logical templates comprising templates of a given utterance and utterances which are equivalent from the point of view of the dialogue controller for a particular part of the dialogue. A logical template CONFIRM may include templates for «Yes», spoken by different speakers and also templates for «Yeah», «Confirm» and other synonyms. The dialogue controller does not need to know which template has been matched: the logical template CONFIRM is identified and the appropriate classified information is sent to the dialogue controller. In a telephone information service, the apparatus includes a speech synthesiser (13), a telephone monitor interface (17) and a data store.
Inventors
- 👤 David Lynton Gibson
Bibliographic Details
Application Number
—
Jurisdiction
EP
Data Source
google_patents
Last Synchronized
2026-10-29 15:02
CPC Classifications
G; G10; G10L; G10L15/00; G10L15/22
Claims Analysis
Total Claims: 15
Claim 1
Independent Claim
132 words
A method of obtaining classified information from utterances spoken by a user in a dialogue controlled by a dialogue controller, comprising establishing templates each representing an utterance in a selectee vocabulary, ano comparing said templates with signals derived from one or more user utterances to determine any close match, characterised by grouping together one or more templates of utterances having the same meaning and/or having equivalent meanings for the purposes of the dialogue such that at least one group comprises templates having equivalent meanings, comparing templates from at least one of said groups with signals derived from one or more utterances spoken in a user response to determine any close match, and sending classified information relating to the identity of the or each group comprising a matched template to the dialogue controller.
Claim 2
Independent Claim
132 words
A method of obtaining classified information from utterances spoken by a user in a dialogue controlled by a dialogue controller, comprising establishing templates each representing an utterance in a selectee vocabulary, ano comparing said templates with signals derived from one or more user utterances to determine any close match, characterised by grouping together one or more templates of utterances having the same meaning and/or having equivalent meanings for the purposes of the dialogue such that at least one group comprises templates having equivalent meanings, comparing templates from at least one of said groups with signals derived from one or more utterances spoken in a user response to determine any close match, and sending classified information relating to the identity of the or each group comprising a matched template to the dialogue controller.
Claim 3
Dependent Claim
41 words
A method as claimed in claim 1, wherein groups are created for stages in possible dialogues controlled by the dialogue controller, wherein a group created for a particular stage or stages comprises templates of utterances having equivalent meanings at said stage.
Claim 4
Dependent Claim
30 words
A method as claimed in Claim 1 or Claim 2, wherein one or more of said groups comprises a plurality of templates of a given utterance spoken by different speakers.
Claim 5
Dependent Claim
27 words
4 . A method as claimed in Claim 3, whereby different speakers establish a template for a given utterance, closely similar templates being discarded prior to grouping.
Claim 6
Independent Claim
32 words
A method as claimec in any preceding claim, wherein one or more of said groups comprises a plurality of templates of a given utterance spoken witr. respective different articulations by one speaker.
Claim 7
Independent Claim
26 words
A method as claimed in any preceding claim, including dynamically adjusting the discrimination threshold for identifying a match according to earlier recognition scores during that dialogue.
Claim 8
Independent Claim
39 words
A method as claimed in any preceding claim, including sending a direction from the dialogue controller to a speech recogniser such that one or more selected groups of templates are ready for comparison with signals derived from said utterances.
Claim 9
Dependent Claim
28 words
A method as claimed in Claim 7, wherein the direction from the dialogue controller causes groups of templates to be downloaded from a template store to the recogniser.
Claim 10
Dependent Claim
38 words
A method as claimed in Claim 7, wherein templates are stored in the speech recogniser and a direction from the dialogue controller causes selected groups of said templates to be activated for a forthcoming part of the dialogue.
Claim 11
Dependent Claim
79 words
A method as claimed in claim 7, including sending directions from the dialogue controller to the speech recogniser for respective forthcoming parts of the dialogue and sending classified information from the speech recogniser to the dialogue controller relating to the or each group of templates comprising a matched template, thereby assimilating information necessary to establish a user enquiry, consulting a data store to obtain information necessary to respona to the enquiry and transmitting the response information to the user.
Claim 12
Dependent Claim
126 words
Apparatus for obtaining classified information from utterances spoken by a user in a dialogue comprising a dialogue controller arranged tc contrcl the cialogue and to communicate with a speech recogniser, characterised in that the dialogue controller is arrangec to group together one or more tenplates representing utterances having the same meaning or equivalent meanings for the purposes of the dialogue where at least one group comprises templates having equivalent meanings, and the speech recogniser is adapted to compare templates within one or more of said groups with signals derived from one or more utterances spoken in a user response to identify any close match, and to send classified information relating to the identity of the or each group comprising a matched template to the dialogue controller.
Claim 13
Dependent Claim
126 words
Apparatus for obtaining classified information from utterances spoken by a user in a dialogue comprising a dialogue controller arranged tc contrcl the cialogue and to communicate with a speech recogniser, characterised in that the dialogue controller is arrangec to group together one or more tenplates representing utterances having the same meaning or equivalent meanings for the purposes of the dialogue where at least one group comprises templates having equivalent meanings, and the speech recogniser is adapted to compare templates within one or more of said groups with signals derived from one or more utterances spoken in a user response to identify any close match, and to send classified information relating to the identity of the or each group comprising a matched template to the dialogue controller.
Claim 14
Dependent Claim
39 words
Apparatus as claimed in claim 11, including a data store for consultation by the dialogue controller to obtain information to answer a user enquiry and transmission means for communicating with the user during the dialogue and transmitting said information.
Claim 15
Dependent Claim
14 words
Apparatus as claimed in claim 12, wherein said transmission means comprises a speech synthesiser.
References Cited (Backward Citations)
FR2533513A1
Cited
USPTO Annuity & Maintenance Windows
Maintenance schedule is calculated for granted US utility patents.