Method and device for recognizing speech in a spelling mode including word qualifiers

• US 5,987,410 A
• Filed: 11/10/1997
• Issued: 11/16/1999
• Est. Priority Date: 11/10/1997
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Inventors: Seide, Frank, Kellner, Andreas
• Primary Examiner(s): Dorvil, Richemond
• Application Number:
  US08/966,722
• Time in Patent Office: 736 Days
• Field of Search: 704/235, 704/231, 704/256, 704/270, 704/240, 704/255
• US Class Current: 704/255
• 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

A method and device for recognizing speech that has a sequence of words each including one or more letters. The word and letters form a recognition data base. The method receives and recognizes the speech by preliminary modelling among various probably recognized sequences. The method selects one or more model sequences as result. In particular, the method allows in a model sequence of exclusively letters, various words as a subset. Such words are used to qualify one or more neighbouring or included letters in the sequence. An applicable model is a mixed information unit model.

Claims

1. A device for recognizing speech units included in a recognition data base, the speech units being a sequence of words having one or more letters, comprisinq:

a speech recognizer for processing the speech units as received and performing preliminary recognition of the speech units;

a spelling filter for generating a number of possible recognized sequences of the speech units, selecting one or more of the possible recognized sequences as a potential recognition result and joining an associated probability score to each of the possible recognized sequences selected;

wherein each of the number of possible recognized sequences of the speech units include a sequence of letters and a subset of one or more of the words in the data base, the subset of the words being a qualifier for one or more the sequence of letters.

2. A method for recognizing speech units included in a recognition data base, the speech units being a sequence of words having one or more letters, said method comprising the steps of:

processing the speech units as received;

performing preliminary recognition of the speech units;

generating a number of possible recognized sequences of the speech units; and

selecting one or more of the possible recognized sequences as a potential recognition result and joining with an associated probability score;

wherein each of the number of possible recognized sequences of the speech units include a sequence of letters and a subset of one or more of the words in the data base, the subset of the words being a qualifier for one or more the sequence of letters.

3. The method as claimed in claim 2, wherein the qualifier implies supplementing any of a diacritical sign, a letter-modifying sign, a doubling indication, or a reference to a whole containing the second level information unit or units in question.

4. The method as claimed in claim 2, being arranged as method for recognizing a word comprising the letters used for a mixed level model.

5. The method as claimed in claim 2, which is configured as a step in a dialog procedure.

6. The method as claimed in claim 5, wherein the step in the dialogue procedure recognizes a single word.