Method for improving acoustic model, computer for improving acoustic model and computer program thereof

A computer for improving an acoustic model, comprising:

• a standard deviation value calculating unit for calculating standard deviation of a feature from first training data to acquire a first standard deviation value and calculating standard deviation of a feature from second training data acquired in a different environment from an environment of the first training data to acquire a second standard deviation value, and wherein environment includes the recording hardware and software used to acquire training data;
• a feature creating unit for creating a feature adapted to an environment where the first training data is recorded, by multiplying the feature acquired from the second training data by a ratio obtained by dividing the first standard deviation value by the second standard deviation value;
• an acoustic model reconstructing unit for reconstructing an acoustic model constructed using training data acquired in the same environment as the environment of the first training data, using the feature adapted to the environment where the first training data is recorded;
• and
• an FMLLR applying unit for creating a feature by applying feature space maximum likelihood linear regression (FMLLR) to the feature acquired from the second training data at the standard deviation value calculating unit, using the acoustic model constructed using training data acquired in the same environment as the environment of the first training data,
• wherein the feature creating unit creates the feature adapted to the environment where the first training data is recorded, by multiplying the feature created by applying the FMLLR at the FMLLR applying unit by the ratio obtained by dividing the first standard deviation value by the second standard deviation value, and
• the acoustic model reconstructing unit reconstructs the acoustic model constructed using the training data acquired in the same environment as the environment of the first training data, using the feature adapted to the environment where the first training data is recorded.

A computer for improving an acoustic model, comprising:

• a standard deviation value calculating unit for calculating standard deviation of a feature from first training data to acquire a first standard deviation value and calculating standard deviation of a feature from second training data acquired in a different environment from an environment of the first training data to acquire a second standard deviation value, and wherein environment includes the recording hardware and software used to acquire training data;
• a feature creating unit for creating a feature adapted to an environment where the first training data is recorded, by multiplying the feature acquired from the second training data by a ratio obtained by dividing the first standard deviation value by the second standard deviation value;
• an acoustic model reconstructing unit for reconstructing an acoustic model constructed using training data acquired in the same environment as the environment of the first training data, using the feature adapted to the environment where the first training data is recorded;
• and
• an FMLLR applying unit for creating a first feature by applying feature space maximum likelihood linear regression (FMLLR) to the feature acquired from the second training data at the standard deviation value calculating unit, using the acoustic model constructed using the training data acquired in the same environment as the environment of the first training data,
• wherein the feature creating unit creates the feature adapted to the environment where the first training data is recorded, by multiplying the first feature created by applying the FMLLR at the FMLLR applying unit by the ratio obtained by dividing the first standard deviation value by the second standard deviation value,
• the FMLLR applying unit further creates a second feature by applying FMLLR to the feature created at the feature creating unit, using the acoustic model constructed using the training data acquired in the same environment as the environment of the first training data, and
• the acoustic model reconstructing unit reconstructs the acoustic model constructed using the training data acquired in the same environment as the environment of the first training data, using the second feature created by applying the FMLLR at the FMLLR applying unit.