Speech recognition device to mark parts of a recognized text

• US 7,376,560 B2
• Filed: 10/09/2002
• Issued: 05/20/2008
• Est. Priority Date: 10/12/2001
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Koninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
• Inventors: Bartosik, Heinrich, Rajic, Kresimir
• Primary Examiner(s): CHAWAN, VIJAY B
• Application Number:
  US10/267,293
• Publication Number:
  US 20030074195A1
• Time in Patent Office: 2,050 Days
• Field of Search: 704/270, 704/260, 704/258, 704/231, 704/251, 704/235, 704/9
• US Class Current: 704/235
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/0631 Creating reference templates; Clustering
• CPC Class Codes: G10L 2015/225 Feedback of the input speech

Abstract

In a transcription device (1) for transcribing a spoken text (GT) into a recognized text (ET) and for editing incorrectly recognized parts of the recognized text (ET), marking means (12, 15, 17) are provided that are arranged for the partly automatic and partly manual marking of parts of the spoken text (GT) and/or of the recognized text (ET) that have a common characteristic. As a result, subsequent unified processing of marked parts of the text that have common characteristics becomes possible.

Claims

1. A transcription device (1) for transcribing a spoken text (GT) into a recognized text (ET) and for editing incorrectly recognized parts of the recognized text (ET), wherein link information (LI) for each part of the spoken text (GT) flags the associated recognized text (ET),

having speech recognition means (2) for transcribing the received spoken text (GT) into the recognized text (ET) and for producing the link information (LI), and having memory means (8) for storing the spoken text (GT), the recognized text (ET) and the link information (LI) and

having marking means (12, 15, 17) for automatically determining marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI) and for storing the marking information determined (CMI, PMI, HMI, GMI, DMI, RMI, WMI) in the memory means (8), which marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI) flags parts of the recognized text (ET) and/or of the spoken text (GT) that have a particular characteristic common to all the marked parts of the text, wherein the marking means (12, 15, 17) have input means (15, 17) for the manual entry of the marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI) to allow further parts of the recognized or spoken text that also have this common characteristic to be marked with this marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI), and

having processing means (13) for unified processing of marked parts of the text.

2. A transcription device (1) as claimed in claim 1, wherein, to allow parameters (PI) of the speech recognition means (2) to be adapted, an adapting stage (11) is provided that is arranged for the evaluation of the automatically and manually marked parts of the text, which marked parts of the text have a common characteristic.

3. A transcription device (1) as claimed in claim 1, wherein reproduction means (13) are arranged for the acoustic reproduction of the spoken text (GT) and the synchronous visual marking of the associated recognized text (ET) when a synchronous reproduction mode is activated in the transcription device (1), which text (ET) is flagged by the link information (LI) and optically reproduced, and wherein the acoustic or optical reproduction of automatically or manually marked parts of the text can be suppressed.

4. A transcription device (1) as claimed in claim 3, wherein the marking means (12, 15, 17) are arranged for the automatic determination and storage of reproduction marking information (WMI) in the memory means (8), which reproduction marking information (WMI) flags those parts of the recognized text (ET) and/or of the spoken text (GT) that were reproduced at least once by the reproduction means (13) when the synchronous reproduction mode was activated.

5. A transcription device (1) as claimed in claim 3, wherein the reproduction means (13) are arranged for the optical reproduction of a replacement text for suppressed marked parts of the recognized text (ET).

6. A transcription device (1) as claimed in claim 5, wherein the reproduction means (13) can switch between optical display of the replacement text and of the marked parts of the text.

7. A transcription device (1) as claimed in claim 6, wherein an additional switch can be made, for marked parts of the text that form text modules, to an optical display of the text defined for the text module.

8. A transcription device (1) as claimed in claim 1, wherein the marking means (12, 15, 17) are arranged for marking a part of a spoken or recognized text with at least two different items of marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI) when this marked part of the text has at least the two common characteristics that are flagged by the items of marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI).

9. A method of transcribing a spoken text (GT) into a recognized text (ET) and of editing incorrectly recognized parts of the recognized text (ET), wherein link information (LI) for each part of the spoken text (GT) flags the associated recognized text (ET) and wherein the following steps are performed:

transcription of the received spoken text (GT) into the recognized text (ET) and production of the link information (LI);

storage of the spoken text (GT), the recognized text (ET) and the link information (LI);

automatic determination of marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI) and storage of the marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI) determined, which marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI) flags parts of the recognized text (ET), and/or spoken text (GT) that have a particular characteristic common to all the marked parts of the text;

manual entry of the marking information (CMI, PMI HMI, GMI, DMI, RMI, WMI) to allow further parts of the recognized or spoken text that also have this common characteristic to be marked with this marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI); and

unified processing of marked parts of the text.

10. A method of transcription as claimed in claim 9, wherein the following further steps are performed:

when a synchronous reproduction mode is activated, acoustic reproduction of the spoken text (GT) and synchronous visual marking of the associated recognized text (ET) that is flagged by the link information (LI) and optically reproduced while the acoustic or optical reproduction of automatically or manually marked parts of the text can be suppressed.

11. A method of transcription as claimed in claim 9, wherein the following further steps are performed:

automatic determination and storage of reproduction marking information (WMI), which reproduction marking information (WMI) flags those parts of the recognized text (ET) and/or of the spoken text (GT) that were reproduced at least once before in the activated synchronous reproduction mode.

12. A method of transcription as claimed in claim 9, wherein the following further steps are performed:

marking of a part of the spoken or recognized text with at least two different items of marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI) when this marked part of the text has at least the two common properties that are flagged by the items of marking information (CMI, PMI, HMI, GMI, DMI, RMI, WMI).