Method and system for searching for an optimal codevector
First Claim
1. A method for searching for an optimal codevector from a plurality of codevectors in a codebook for data modeling of input signals to a data processing system, including a system for processing speech signals, the optimal codevector having a minimum weighted distance to a given vector, the codevectors comprising a plurality of components, the method comprising:
- determining, for an input signal to a processing system to convert the input signal from an analog representation to a digital representation, a partial distance between a current vector component of a current codevector and of a corresponding component of the given vector;
comparing the partial distance to a saved renormalized minimum partial distance; and
proceeding to a next codevector when the saved renormalized minimum partial distance is smaller than the partial distance, whereby an optimal representation of the input signal as a digital signal in the processing system is efficiently achieved.
2 Assignments
0 Petitions
Accused Products
Abstract
Method and system aspects for searching for an optimal codevector from a plurality of codevectors in a codebook, the optimal codevector having a minimum distance to a given vector, are provided. The aspects determine a partial distance with a current vector component of a current codevector and of the given vector, compare the partial distance to a saved renormalized minimum partial distance, and proceed to a next codevector when the saved renormalized minimum partial distance is smaller than the partial distance.
In addition, the present invention proceeds to a next vector component when the partial distance is smaller than the saved renormalized minimum partial distance. When the partial distance computed with each next vector component is smaller than the saved renormalized minimum partial distance, the present invention calculates a full weighted distance value, compares it to a saved minimum full weighted distance, and updates chosen values. The operation then continues with a next codevector until all codevectors have been used. An optimal index to identify the optimal codevector is returned when all codevectors have been used.
30 Citations
25 Claims
-
1. A method for searching for an optimal codevector from a plurality of codevectors in a codebook for data modeling of input signals to a data processing system, including a system for processing speech signals, the optimal codevector having a minimum weighted distance to a given vector, the codevectors comprising a plurality of components, the method comprising:
-
determining, for an input signal to a processing system to convert the input signal from an analog representation to a digital representation, a partial distance between a current vector component of a current codevector and of a corresponding component of the given vector;
comparing the partial distance to a saved renormalized minimum partial distance; and
proceeding to a next codevector when the saved renormalized minimum partial distance is smaller than the partial distance, whereby an optimal representation of the input signal as a digital signal in the processing system is efficiently achieved. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10)
-
-
11. A system for searching for an optimal codevector from a plurality of codevectors in a codebook for data modeling of input signals to the system, including speech signals, the optimal codevector having a minimum distance to a given vector, the codevectors comprising a plurality of components, the system comprising:
-
storage medium storing the codebook; and
a central processing unit (CPU) coupled to the storage medium and capable of coordinating the searching including determining, for an input signal to the system to convert the input signal from an analog representation to a digital representation, a partial distance between a current vector component of a current codevector and of a corresponding component of the given vector, comparing the partial distance to a saved renormalized minimum partial distance, and proceeding to a next codevector when the saved renormalized minimum partial distance is smaller than the partial distance, whereby an optimal representation of the input signal as a digital signal in the system is efficiently achieved.
-
-
12. The system of 11 wherein the CPU further coordinates proceeding to a next vector component when the partial distance is smaller than the saved renormalized minimum partial distance.
-
13. The system of 12 wherein the CPU further coordinates calculating a full weighted distance value by summing together the partial distance computed with each next vector component after multiplication with corresponding weights when the partial distance computed with each next vector component is smaller than the saved renormalized minimum partial distance.
-
14. The system of 13 wherein the CPU further coordinates comparing the full weighted distance to a saved minimum full weighted distance.
-
15. The system of 14 wherein the CPU further coordinates updating an optimal index to an index of the current codevector, updating the saved minimum full weighted distance to the full weighted distance, and updating the saved renormalized minimum partial distance when the full weighted distance is smaller than the saved minimum full weighted distance.
-
16. The system of 15 wherein the CPU further coordinates continuing with a next codevector until all of the plurality of codevectors have been used.
-
17. The system of 14 wherein the CPU further coordinates continuing with a next codevector when the saved minimum full weighted distance is smaller than the full distance until all of the plurality of codevectors have been used.
-
18. The system of 16 wherein the CPU further coordinates returning the optimal index to identify the optimal codevector when all of the plurality of codevectors in the codebook have been used.
-
19. A system for searching for an optimal codevector from a plurality of codevectors in a codebook, for data modeling of input signals to the system, including speech signals, the optimal codevector having a minimum distance to a given vector, the codevectors comprising a plurality of components, the system comprising:
-
means for storing the codebook; and
means for utilizing the codebook, the means for utilizing coupled to the storing means and coordinating the searching, including determining, for an input signal to the system to convert the input signal from an analog representation to a digital representation, a partial distance between a current vector component of a current codevector and of a corresponding component of the given vector, comparing the partial distance to a saved renormalized minimum partial distance, and proceeding to a next codevector when the saved renormalized minimum partial distance is smaller than the partial distance, whereby an optimal representation of the input signal as a digital signal in the system is efficiently achieved.
-
-
20. The system of 19 wherein the means for utilizing further coordinates proceeding to a next vector component when the partial distance is smaller than the saved renormalized minimum partial distance.
-
21. The system of 20 wherein the means for utilizing further coordinates calculating a full weighted distance value by summing together the partial distance computed with each next vector component after multiplication with corresponding weights when the partial distance computed with each next vector component is smaller than the saved renormalized minimum partial distance, and comparing the full weighted distance to a saved minimum full weighted distance.
-
22. The system of 21 wherein the means for utilizing further coordinates updating an optimal index to an index of the current codevector, updating the saved minimum full weighted distance to the full weighted distance, and updating the saved renormalized minimum partial distance when the full weighted distance is smaller than the saved minimum full weighted distance, and continuing with a next codevector until all of the plurality of codevectors have been used.
-
23. The system of 22 wherein the means for utilizing further coordinates returning the optimal index to identify the optimal codevector when all of the plurality of codevectors in the codebook have been used.
-
24. A computer readable medium containing program instructions for searching for an optimal codevector from a plurality of codevectors in a codebook for data modeling of input signals to a data processing system, including a system for processing speech signals, the optimal codevector having a minimum distance to a given vector, the codevectors comprising a plurality of components, the program instructions comprising:
-
determining, for an input signal to a processing system, a partial distance with a current vector component of a current codevector and of the given vector;
comparing the partial distance to a saved renormalized minimum partial distance;
proceeding to a next codevector when the saved renormalized minimum partial distance is smaller than the partial distance; and
proceeding to a next vector component when the partial distance is smaller than the saved renormalized minimum partial distance, whereby an optimal representation of the input signal in the processing system is efficiently achieved.
-
-
25. A computer readable medium containing program instructions for searching for an optimal codevector from a plurality of codevectors in a codebook for data modeling of input signals to a data processing system, including a system for processing speech signals, the optimal codevector having a minimum distance to a given vector, the codevectors comprising a plurality of components, the program instructions comprising:
-
determining, for an input signal to a processing system to convert the input signal from an analog representation to a digital representation, a partial distance with a current vector component of a current codevector and of the given vector;
comparing the partial distance to a saved renormalized minimum partial distance;
proceeding to a next codevector when the saved renormalized minimum partial distance is smaller than the partial distance; and
proceeding to a next vector component when the partial distance is smaller than the saved renormalized minimum partial distance, whereby an optimal representation of the input signal as a digital signal in the processing system is efficiently achieved.
-
Specification