US2009326932A1PendingUtilityA1
Reducing Computational Complexity in Determining the Distance from Each of a Set of Input Points to Each of a Set of Fixed Points
Est. expiryAug 18, 2025(expired)· nominal 20-yr term from priority
Inventors:Chanaveeragouda V. Goudar
G06F 7/552G06F 2207/5525G10L 15/10
50
PatentIndex Score
0
Cited by
0
References
0
Claims
Abstract
An aspect of the present invention takes advantage of the fact that the coordinates of fixed points do not change, and thus the energy (sum of squares of the coordinates defining the vector) of each fixed point is computed and stored. The energy of each variable input point may also be computed. The distance between each pair of fixed and input points is computed based on the respective energies and the dot product.
Claims
exact text as granted — not AI-modified1 - 3 . (canceled)
4 . A method for encoding in a speech encoder, said method comprising:
receiving a stream of uncompressed digital codes; computing a weighted distance between each of L input points and R fixed points according to a weight vector, wherein L and R are integers having a value greater than 1, the with wth one of said input points being represented by (Iw1, Iw2, . . . , Iwn) and jth one of said fixed points being represented by (Fj(1), Fj(2), . . . , Fj(n)), said weighted vector being represented by (G(1), G(2), . . . , G(n)), and said weighted distance (WMSEw,j) between said wth input point and said jth fixed point being represented by:
WMSE
w
,
j
=
∑
i
=
l
n
G
(
i
)
(
I
w
(
i
)
-
F
j
(
i
)
)
2
said computing comprising:
computing a first aggregated value equaling:
∑
i
=
1
10
W
LSF
(
i
)
(
I
LSF
l
(
i
)
)
2
computing a second aggregated value equaling
∑
i
=
1
10
W
LSF
(
i
)
(
C
LSF
r
(
i
)
)
2
computing a third aggregated value equaling
∑
i
=
1
n
(
-
2
*
G
(
i
)
I
w
(
i
)
)
·
F
j
(
i
)
and adding said first aggregated value, said second aggregated value and said third aggregated value to generate said weighted distance;
and
encoding said uncompressed digital codes into a compressed format using the results of computing the distance between each of L input points and R fixed points.
5 . The method of claim 4 , wherein said L input points represent L samples of speech, and R fixed points represents entries a code book according to which said L samples are represented in a compressed format.
6 . The method of claim 5 , wherein compressed format is according to 3 Gpp2 standard.
7 . The method of claim 5 , wherein said weighted distance is computed for each combination of j=1 to R and I=1 to L.
8 . A computer readable medium carrying one or more sequences of instructions for causing a system to compute the weighted distance between each of L input points and R fixed points according to a weight vector, wherein L and R are integers having a value greater than 1, the with wth one of said input points being represented by (Iw1, Iw2, . . . , Iwn) and jth one of said fixed points being represented by (Fj(1), Fj(2), . . . , Fj(n)), said weighted vector being represented by (G(1), G(2), . . . , G(n)), and said weighted distance (WMSEw,j) between said wth input point and said jth fixed point being represented by:
wherein execution of said one or more sequences of instructions by one or more processors contained in said system causes said one or more processors to perform the actions of:
computing a first aggregated value equaling:
∑
i
=
1
10
W
LSF
(
i
)
(
I
LSF
l
(
i
)
)
2
computing a second aggregated value equaling
∑
i
=
1
10
W
LSF
(
i
)
(
C
LSF
r
(
i
)
)
2
computing a third aggregated value equaling
∑
i
=
1
n
(
-
2
*
G
(
i
)
I
w
(
i
)
)
·
F
j
(
i
)
and adding said first aggregated value, said second aggregated value and said third aggregated value to generate said weighted distance.
9 . The computer readable medium of claim 8 , wherein said L input points represent L samples of speech, and R fixed points represents entries a code book according to which said L samples are represented in a compressed format.
10 . The computer readable medium of claim 9 , wherein compressed format is according to 3 Gpp2 standard.
11 - 23 . (canceled)Join the waitlist — get patent alerts
Track US2009326932A1 — get alerts on status changes and closely related new filings.
We store only your email — no account needed. See our privacy policy.