Method for determining the presence of a wanted signal component
View Patent ↗This invention provides a method for determining, in a speech dialog system issuing speech prompts, a score value as an indicator for the presence of a wanted signal component in an input signal stemming from a microphone, comprising the steps of: using a first likelihood function to determine a first likelihood value for the presence of the wanted signal component in the input signal, using a second likelihood function to determine a second likelihood value for the presence of a noise signal component in the input signal, and determining a score value based on the first and the second likelihood values, wherein the first likelihood function is based on a predetermined reference wanted signal, and the second likelihood function is based on a predetermined reference noise signal.
1. A method for determining, in a speech dialogue system issuing speech prompts, a score value as an indicator for the presence of a wanted signal component in an input signal stemming from a microphone, comprising the steps of:
using a first likelihood function to determine a first likelihood value for the presence of the wanted signal component in the input signal,
using a second likelihood function to determine a second likelihood value for the presence of a noise signal component in the input signal, and
determining, using a processor, a score value based on the first and the second likelihood values, wherein the first likelihood function is based on a predetermined reference wanted signal, and the second likelihood function is based on a predetermined reference noise signal,
wherein the redetermined reference wanted signal is based on a first superposition signal obtained by superimposing at least one model wanted signal on at least one model prompt signal, and the predetermined reference noise signal is based on a second superposition signal obtained by superimposing at least one model noise signal on the at least one model prompt signal.
2. The method of claim 1 , wherein the first likelihood function is based on parameters derived from the first superposition signal, and the second likelihood function is based on parameters derived from the second superposition signal.
3. The method of claim 2 , wherein the parameters determine properties of mathematical functions, in particular, of multivariate Gaussian distribution functions.
4. The method according to claim 1 , wherein the first likelihood value and the second likelihood value are determined as a function of a first and a second reference superposition signal, respectively.
5. The method according to claim 1 , further comprising the step of selecting the first likelihood function and/or the second likelihood function from a plurality of candidate likelihood functions depending on the environment of the microphone.
6. The method according to claim 1 , further comprising the step of selecting the first likelihood function or the second likelihood function from a plurality of candidate likelihood functions depending on a particular user of the speech dialogue system.
7. The method according to claim 1 , wherein the first likelihood function and/or the second likelihood function are time dependent.
8. The method of according to claim 1 , further comprising the step of:
using a third likelihood function to determine a third likelihood value for the presence of a background speech signal component in the input signal; wherein the third likelihood function is based on a predetermined reference background speech signal.
9. The method of claim 8 , wherein the predetermined reference background speech signal is based on a third superposition signal obtained by superimposing at least one model background speech signal on at least one model prompt signal.
10. The method according to claim 1 , further comprising reducing noise in the input signal before determining at least one of the likelihood values.
11. The method according to claim 1 , further comprising reducing noise in the input signal after determining at least one of the likelihood values.
12. The method according to claim 1 , wherein the score value is based on the first likelihood value divided by a sum of the likelihood values.
13. The method according to claim 1 , further comprising the step of:
comparing the score value with a threshold value.
14. The method of claim 1 , further including transforming the wanted signal component from a user speech command to a system instruction.
15. A computer program product comprising one or more non-transitory computer readable media having computer code stored thereon for producing a score value as an indicator for the presence of a wanted signal component in an input signal stemming from a microphone, the computer code comprising:
computer code using a first likelihood function to determine a first likelihood value for the presence of the wanted signal component in the input signal,
computer code using a second likelihood function to determine a second likelihood value for the presence of a noise signal component in the input signal, and
computer code for determining a score value based on the first and the second likelihood values, wherein the first likelihood function is based on a predetermined reference wanted signal, and the second likelihood function is based on a predetermined reference noise signal,
wherein the predetermined reference wanted signal is based on a first superposition signal obtained by superimposing at least one model wanted signal on at least one model prompt signal, and the predetermined reference noise signal is based on a second superposition signal obtained by superimposing at least one model noise signal on the at least one model prompt signal.
16. The computer program product of claim 15 , wherein the first likelihood function is based on parameters derived from the first superposition signal, and the second likelihood function is based on parameters derived from the second superposition signal.
17. The computer program product of claim 16 , wherein the parameters determine properties of mathematical functions, in particular, of multivariate Gaussian distribution functions.
18. The computer program product according to claim 15 , wherein the first likelihood value and the second likelihood value are determined as a function of a first and a second reference superposition signal, respectively.
19. The computer program product according to claim 15 , further comprising:
computer code for selecting the first likelihood function and/or the second likelihood function from a plurality of candidate likelihood functions depending on the environment of the microphone.
20. The computer program product according to claim 15 , further comprising:
computer code for selecting the first likelihood function or the second likelihood function from a plurality of candidate likelihood functions depending on a particular user of the speech dialogue system.
21. The computer program product according to claim 15 , wherein the first likelihood function and/or the second likelihood function are time dependent.
22. The computer program product of according to claim 15 , further comprising:
computer code for using a third likelihood function to determine a third likelihood value for the presence of a background speech signal component in the input signal; wherein the third likelihood function is based on a predetermined reference background speech signal.
23. The computer program product of claim 22 , wherein the predetermined reference background speech signal is based on a third superposition signal obtained by superimposing at least one model background speech signal on at least one model prompt signal.
24. The computer program product according to claim 15 , further comprising:
computer code for reducing noise in the input signal before determining at least one of the likelihood values.
25. The computer program product according to claim 15 , further comprising:
computer code for reducing noise in the input signal after determining at least one of the likelihood values.
26. The computer program product according to claim 15 , wherein the score value is based on the first likelihood value divided by a sum of the likelihood values.
27. The computer program product according to claim 15 , further comprising:
computer code for comparing the score value with a threshold value.
28. An apparatus which is adapted to determine, in a speech dialogue system issuing speech prompts, a score value as an indicator for the presence of a wanted signal component in an input signal stemming from a microphone, comprising the components of:
means for determining a first likelihood value for the presence of the wanted signal component in the input signal by using a first likelihood function,
means for determining a second likelihood value for the presence of a noise signal component in the input signal by using a second likelihood function, and
means for determining a score value based on the first and the second likelihood values, wherein the first likelihood function is based on a predetermined reference wanted signal, and the second likelihood function is based on a predetermined reference noise signal,
wherein the predetermined reference wanted signal is based on a first superposition signal obtained by superimposing at least one model wanted signal on at least one model prompt signal, and the predetermined reference noise signal is based on a second superposition signal obtained by superimposing at least one model noise signal on the at least one model prompt signal.