Pronunciation measurement device and method
First Claim
1. A method of measuring pronunciation, comprising:
- receiving voice input and processing the voice input to provide a plurality of voice input phonemes;
performing a look-up operation to obtain a predetermined model for the voice input, wherein the predetermined model comprises a plurality of model phonemes;
applying the voice input to the model by comparing the voice input phonemes with the model phonemes to provide a score;
analyzing the score with respect to a score for a predetermined speaker, including comparing a duration of at least one voice input phoneme with a duration of at least one model phoneme, thereby providing a result; and
indicating the result including indicating a confidence measure for the duration of the at least one voice input phoneme.
4 Assignments
0 Petitions
Accused Products
Abstract
Upon selection of an expression for pronunciation training, a look-up operation is performed in a speaker database (15) to obtain a predetermined model for comparison with a voice of a user received at an input (11). A speech modeling element models speech of a native speaker. The voice input is applied to the modeling element (102-107) and an analysis is carried out of the comparison, in correlation and in duration, between a phoneme or sub-word of the input and a phoneme or sub-word of the native speaker to provide a score, including a score for the correlation and a score for the duration. The score is analyzed with respect to a score for a predetermined speaker in an analysis element (40). An indicator device (16) coupled to the output of the analysis element indicates the result in a graphical illustration. A tracking tool indicates state of progress of the voice of the speaker.
66 Citations
16 Claims
-
1. A method of measuring pronunciation, comprising:
-
receiving voice input and processing the voice input to provide a plurality of voice input phonemes; performing a look-up operation to obtain a predetermined model for the voice input, wherein the predetermined model comprises a plurality of model phonemes; applying the voice input to the model by comparing the voice input phonemes with the model phonemes to provide a score; analyzing the score with respect to a score for a predetermined speaker, including comparing a duration of at least one voice input phoneme with a duration of at least one model phoneme, thereby providing a result; and indicating the result including indicating a confidence measure for the duration of the at least one voice input phoneme. - View Dependent Claims (2, 3, 6, 7)
-
-
4. A method of measuring pronunciation, comprising:
-
receiving voice input; performing a look-up operation to obtain a predetermined model for the voice input; applying the voice input to the model to provide a score, including providing a first output providing measurements of durations of sub-words and a second output providing measurements of correlations between sub-words in the voice input and sub-words in the predetermined model; analyzing the score with respect to a score for a predetermined speaker, thereby providing a result wherein the step of analyzing the score comprises performing statistical analysis of the first output with respect to predetermined measurements of durations of sub-words for the predetermined speaker; and indicating the result, including indicating a confidence measure for durations of sub-words in the voice input. - View Dependent Claims (5)
-
-
8. A device for pronunciation measurement comprising:
-
a speech modeling element having an input to receive a signal representing a voice of a speaker and an output; a speaker database; an analysis element having a first input coupled to the speaker database and a second input coupled to the output of the speech modeling element and having an output; and a graphic user interface indicator device coupled to the output of the analysis element, including an indicator of confidence measure for a duration of a phoneme of the voice of the speaker and an indicator of a confidence measure for quality of a phoneme of the voice of the speaker. - View Dependent Claims (9, 10, 11, 12, 13, 14, 15, 16)
-
Specification