Energy feature extraction method for noisy speech recognition
AssigneeIndustrial Technology Research Institute
InventorTai-Huei Huang
Life of this patent
Estimating the term…
Abstract
The present invention provides an energy feature extraction method for noisy speech recognition. At first, noisy speech energy of an input noisy speech is computed. Next, the noise energy in the input noisy speech is estimated. Then, the estimated noise energy is subtracted from the noisy speech energy to obtain estimated clean speech energy. Finally, delta operations are performed on the log of the estimated clean speech energy to determine the energy derivative features for the noisy speech.
Claims10
An energy feature extraction method for noisy speech recognition, comprising:
- a noisy speech energy computation step for computing a noisy speech energy of an input noisy speech;
- a noise energy estimation step for estimating a noise energy in the input noisy speech through the noisy speech energy and an energy dynamic range ratio which is defined as a ratio of maximum frame energy to minimum frame energy in speech, thereby obtaining an estimated noise energy êw as follows:
- e
- ^
- w
- =
- max
- t
- =
- 1
- ,
- T
-
- (
-
- E
- y
-
- [
- t
- ]
- )
- -
- r
- x
-
- min
- t
- =
- 1
- ,
- T
-
- (
-
- E
- y
-
- [
- t
- ]
- )
- 1
- -
- r
- x
- ,
- where Ey[t] is the noisy speech energy, T is the number of speech frames, rx is the energy dynamic range ratio of clean speech;
- a speech energy estimation step in which the estimated noise energy is subtracted from the noisy speech energy to obtain an estimated clean speech energy;
- an energy feature computation step for performing delta operation on log of the estimated clean speech energy to determine differential coefficient for automatic speech recognition;
- and
- an output step for outputting the differential coefficient determined in the energy feature computation step to perform noisy speech recognition.
An energy feature extraction method for noisy speech recognition, comprising:
- a noisy speech energy computation step for computing a noisy speech energy of an input noisy speech;
- a noise energy estimation step for estimating a noise energy in the input noisy speech;
- a speech energy estimation step in which the estimated noise energy is subtracted from the noisy speech energy to obtain an estimated clean speech energy;
- an energy feature computation step which performs delta operation on log of the estimated clean speech energy to determine differential coefficient for automatic speech recognition;
- and
- an output step for outputting the differential coefficient determined in the energy feature computation step to perform noisy speech recognition;
- wherein, in the noise energy estimation step, the noise energy in the noisy speech is estimated by directly computing the energy of pure background noise.
Dependencies are inferred from the claim text.