Method of training an automatic speech recognizer
AssigneeKoninklijke Philips Electronics N.V.
InventorsGeorg Rose, Joseph Hubertus Eggen, Bartel Marinus Van Der Sluis
Life of this patent
Estimating the term…
Abstract
Provided is a method of training an automatic speech recognizer, said speech recognizer using acoustic models and/or speech models, wherein speech data is collected during a training phase and used to improve the acoustic models, said method comprising: during the training phase, providing speech utterances that are predefined to a user by means of a game, wherein the game has predefined rules to enable a user to provide certain utterances; and providing the utterances by the user for training the speech recognizer.
Claims7
A method of training an automatic speech recognizer, said speech recognizer using acoustic models and/or speech models, wherein speech data is collected during a training phase and used to improve the acoustic models, said method comprising:
- during the training phase, providing speech utterances that are predefined to a user as marks for assigned picture screen areas;
- providing the utterances by the user for training the speech recognizer;
- shifting the marks on the picture screen when the associated speech utterance is made to produce as a target a predefinable structure on the picture screen;
- and
- redefining the mark after the target has been reached.
A speech recognition system for implementing a method as claimed in 1.
A method of speaker adaptation of an automatic speech recognizer, said speech recognizer using acoustic models and/or speech models, wherein speech data is collected and used to improve the acoustic models, said method comprising:
- providing speech utterances that are predefined for a user as marks for assigned picture screen areas;
- providing the utterances for adapting the speech recognizer to the user;
- shifting the marks on the picture screen when the associated speech utterance is made to produce as a target a predefinable structure on the picture screen;
- and
- redefining the mark after the target has been reached.
Dependencies are inferred from the claim text.