Speech recognition system, training arrangement and method of calculating iteration values for free parameters of a maximum-entropy speech model

• US 20020156628A1
• Filed: 02/13/2002
• Published: 10/24/2002
• Est. Priority Date: 02/13/2001
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Koninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
• Inventors: Peters, Jochen
• Granted Patent: US 7,010,486 B2
• Time in Patent Office: Days
• US Class Current: 704/255
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 15/197 Probabilistic grammars, e.g. word n-grams

Abstract

The invention relates to a speech recognition system and a method of calculating iteration values for free parameters λ_α^ortho(n) of a maximum-entropy speech model MESM with the aid of the generalized-iterative scaling training algorithm in a computer-supported speech recognition system in accordance with the formula

λ_α^ortho(n+1)=G(λ_α^ortho(n), m_α^ortho, . . . )

where n is an iteration parameter, G a mathematical function, α an attribute in the MESM and m_α^ortho a desired orthogonalized boundary value in the MESM for the attribute α. It is an object of the invention to further develop the system and method so that they make a fast computation of the free parameters λ possible without a change of the original training object. According to the invention this object is achieved in that the desired orthogonalized boundary value m_α^ortho is calculated by a linear combination of the desired boundary value m_α with desired boundary values m_β from attributes β that have a larger range than the attribute α. m_α and m^β are then desired boundary values of the original training object.

Claims

1. A method of calculating iteration values for free parameters λ

_α

^ortho(n) of a maximum-entropy speech model MESM in a speech recognition system with the aid of the generalized iterative scaling training algorithm in accordance with the following formula;

λ

_α

^{ortho( n+l)}=G(λ

_α

^ortho(n), m_α

^ortho, . . . )

where;

n;

is an iteration parameter;

G;

is a mathematical function;

α

;

is an attribute in the MESM; and

m_α

^ortho;

is a desired orthogonalized boundary value in the MESM for the attribute α

, characterized in that the desired orthogonalized boundary value m_α

^ortho is calculated by linearly combining the desired boundary value m_α

with desired boundary values m_β

of attributes β

that have a larger range than the attribute α

.

2. A method as claimed in claim 1, characterized in that the calculation of the desired orthogonalized boundary value m_α

^ortho for the attribute α

=β

0 comprises the following steps;

a) Selecting all the attributes β

i with i=1. . . g in the speech model that have a larger range RW than the attribute α

=β

0 and include the latter;

b) Calculating desired boundary values mβ

i for the attributes β

i with i=0 . . . g;

c) Sorting the attributes β

i with i=0 . . . g according to their RW;

d) Selecting one of the attributes β

i having the largest RW;

e) Checking whether there are other attributes β

k which include the attribute β

i and have a larger RW than the selected attribute β

i;

f1) If so, defining a parameter X as a linear combination of the orthogonalized boundary values m<sub>β

k</sub>^ortho calculated in step g) during the last run of the steps e) to g) for all the attributes β

k that have a larger range and are determined in the most recently run step e);

f2) If not, defining the parameters X to X=0;

g) Calculating the desired orthogonalized boundary value m<sub>β

k</sub>^ortho for the attribute pi by arithmetically combining the desired boundary value mβ

i with a parameter X; and

h) Repeating the steps e) to g) for the attribute β

i-1 whose RW is smaller than or equal to the RW of the attribute β

i until the desired orthogonalized boundary value m<sub>β

0</sub>^ortho=m_α

^ortho with i=0 has been calculated in step g).

3. A method as claimed in claim 2, characterized in that the calculation of the parameter X in step f1) is made according to the following formula:

4. A method as claimed in claim 3, characterized in that the calculation of the desired orthogonalized boundary value m_β

1^ortho is made in step g) according to the following formula;

5. A method as claimed in claim 2, characterized in that the calculation of the desired boundary values m_β

1 for the attributes β

i with i=0, . . . , g is made in step b) by respectively calculating the frequency N(β

i), with which the attribute β

i occurs in a training corpus and by subsequently smoothing the calculated frequency value N(β

i).

6. A method as claimed in claim 5, characterized in that the calculation of the frequency N(β

i) is made by applying a binary attribute function fβ

i to the training corpus where fβ

i is defined as;

7. A method as claimed in claim 1, characterized in that the mathematical function G has as a further variable the magnitude of a convergence step t_α

^ortho with;

t_α

^ortho=1/M^ortho

where

Mortho;

represents for binary functions ƒ

_α

^ortho the maximum number of functions which yield the value 1 for the same argument (h,w).

8. A method as claimed in claim 7, characterized in that the attribute function ƒ

_α

^ortho is calculated by linearly combining an attribute function ƒ

_α

with orthogonalized attribute functions ƒ

_β

^ortho is calculated from attributes p that have a larger range than the attribute α

.

9. A method as claimed in claim 8, characterized in that the calculation of the orthogonalized attribute function ƒ

_α

^ortho for the attribute α

=β

0 comprises the following steps;

a) Selecting all the attributes β

i with i=1 . . . g in the speech model that have a larger range RW than the attribute α

=β

0 and include the latter;

b) Calculating boundary values fβ

i for the attributes β

i with i=0 . . . g;

c) Sorting the attributes β

i with i=0 . . . g according to their RW;

d) Selecting one of the attributes β

i having the largest RW;

e) Checking whether there are other attributes β

k which include the attribute β

i and have a larger RW than the selected attribute β

i;

f1) If so, defining a function F as a linear combination of the orthogonalized attribute function ƒ

<sub>β

k</sub>^ortho calculated in step g) during the last run of the steps e) to g) for all the attributes β

k that have a larger range determined in the most recently run step e);

f2) If not, defining the function F to F=0;

g) Calculating the orthogonalized attribute function ƒ

<sub>β

k</sub>^ortho for the attribute Pi by arithmetically combining the attribute function fβ

i with the function F; and

h) Repeating the steps e) to g) for the attribute β

i-1 whose range is smaller than or equal to the range of the attribute β

i until the orthogonalized attribute function ƒ

<sub>β

0</sub>^ortho=ƒ

_α

^ortho with i=0 has been calculated in step g).

10. A method as claimed in claim 9, characterized in that the calculation of the function F in step f1) is made according to the following formula:

11. A method as claimed in claim 9, characterized in that the calculation of the orthogonalized attribute function ƒ

<sub>β

i</sub>^ortho in step g) is made according to the following formula;

ƒ

<sub>β

i</sub>^ortho=ƒ

<sub>β

i</sub>−

F

12. A method as claimed in claim 1, characterized in that the mathematical function G has the following form:

13. A method as claimed in claim 1, characterized in that the mathematical function has the following form:

14. A speech recognition system (10) comprising:

a recognition device (12) for recognizing the semantic content of an acoustic signal captured and rendered available by a microphone (20), more particularly a speech signal, by mapping parts of this signal onto predefined recognition symbols as they are offered by the implemented maximum-entropy speech model MESM, and for generating output signals which represent the recognized semantic content; and

a training system (14) for adapting the MESM to recurrent statistical patterns in the speech of a certain user of the speech recognition system (10);

characterized in that the training system (14) calculates free parameters λ

in the MESM in accordance with the method as claimed in claim 1.

15. A training system (14) for adapting the maximum-entropy speech model MESM in a speech recognition system (10) to recurrent statistical patterns in the speech of a certain user of this speech recognition system (10), characterized in that the training system (14) calculates free parameters λ

in the MESM in accordance with the method as claimed in claim 1.