Method of automatic recognition of a spelled speech utterance

• US 6,725,197 B1
• Filed: 10/12/1999
• Issued: 04/20/2004
• Est. Priority Date: 10/14/1998
• Status: Expired due to Fees

• Current Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Original Assignee: Koninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
• Inventors: Wuppermann, Friedhelm, Stahl, Volker
• Primary Examiner(s): {haeck over (S)}mits, Talivaldis Ivars
• Assistant Examiner(s): Azad, Abul K.
• Application Number:
  US09/415,832
• Time in Patent Office: 1,652 Days
• Field of Search: 704/251-255, 704/257
• US Class Current: 704/251
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/086 Recognition of spelled words

Abstract

The invention relates to a method of automatic recognition of an at least partly spelled speech utterance, with a speech recognition unit (2) based on statistical models that include a linguistic speech model (6).

To improve this method in such a way that in addition to a convenient entry, also a reduced speech recognition error rate is achieved, there is proposed that

after the at least partly spelled speech utterance has been entered, the speech recognition unit (2) determines a first recognition result for the whole speech utterance;

individually recognized letters are sent to the user for him to acknowledge or reject;

after a letter has been acknowledged, the linguistic speech model (6) is adapted, which linguistic speech model, after its adaptation, determines the number of letters that can be allowed as followers of the acknowledged letters and assumes the correctness of letters already acknowledged;

the speech recognition unit (2) determines a further recognition result for the speech utterance by means of the adapted linguistic speech model (6), from which result the next letter to be sent to the user is determined, so that he can acknowledge it.

Claims

1. A method of automatic recognition of an at least partly spelled speech utterance, with a speech recognition unit based on statistical models including a linguistic speech model, in which

after the at least partly spelled speech utterance has been entered, the speech recognition unit determines a first recognition result for the speech utterance;

individually recognized letters are sent to a user with a request to acknowledge or reject;

after a letter has been acknowledged the linguistic speech model is adapted, which linguistic speech model determines the number of the letters that can be allowed as followers of the acknowledged letter and assumes the correctness of letters already acknowledged;

with the adapted linguistic speech model the recognition unit determines a further recognition result for the speech utterance from which result the next letter to be sent to the user is determined, so that the user can acknowledge it.

2. A method as claimed in claim 1, characterized in that, after a recognized letter has been rejected, the linguistic speech model is adapted in accordance with this information and the speech recognition unit carries out a new recognition operation.

3. A method as claimed in claim 1, characterized in that the speech recognition unit determines a list N of best recognition alternatives as a recognition result for the speech utterance and in that, after the user has rejected a recognized letter, the user is given the respective letter of the second-best solution alternative.

4. A method as claimed in claim 1, characterized in that separate letter are assigned individual probability vales which particularly depend on all the position-specific previous letters.

5. A method as claimed in claim 1, wherein the step of individually recognized letters further includes, when the user has rejected a letter, determining a recognition alternative to the letter and, the degree of exchangeability with other letters expressed in probability values is taken into account.

6. A method as claimed in claim 1, characterized in that invalid and wrongly recognized initial letter combinations in the linguistic speech model are defined as invalid and are not proposed to the user;

in such a case the speech recognition unit determines a further recognition result for the speech utterance by means of the adapted linguistic speech model from which result is determined the letter to be supplied to the user for him to acknowledge.

7. A method as claimed in claim 5, characterized in that letters for which there are no recognition alternatives are automatically processed as acknowledged letters, even without n acknowledgement by the user.

8. An electrical appliance comprising:

a processor with a speech recognition unit using linguistic speech model,

wherein after an at least partly spelled speech utterance has been entered, the speech recognition unit determines a first recognition result for the speech utterance, and individually recognized letters are sent to a user with a request to acknowledge or reject;

after a letter has been acknowledged the linguistic speech model is adapted;

with the adapted linguistic speech model the speech recognition unit determines a further recognition result for the speech utterance.