Systems and methods of pre-processing of speech signals for improved speech recognition
Pre-processing systems, methods of pre-processing, and speech processing systems for improved Automated Speech Recognition are provided. Some pre-processing systems for improved speech recognition of a speech signal are provided, which systems comprise a pitch estimation circuit; and a pitch equalization processor. The pitch estimation circuit is configured to receive the speech signal to determine a pitch index of the speech signal, and the pitch equalization processor is configured to receive the speech signal and pitch information, to equalize a speech pitch of the speech signal using the pitch information, and to provide a pitch-equalized speech signal.
1. A pre-processing system for improved speech recognition of a speech signal, the system comprising at least
a pitch estimation circuit; and
a pitch equalization processor; wherein
the pitch estimation circuit is configured to receive the speech signal to determine a pitch index of the speech signal; and wherein
the pitch equalization processor is configured to
receive the speech signal and pitch information;
equalize a speech pitch of the speech signal using the pitch information through removal of the speech pitch of the speech signal and re-synthesis of the speech pitch of the speech signal with a predefined speech recognition pitch index; and to
provide a pitch-equalized speech signal.
2. The pre-processing system of claim 1 , further comprising a pitch averaging circuit; which pitch averaging circuit is configured to receive the pitch index from the pitch estimation circuit, to determine an average pitch index, and to provide the pitch information to the pitch equalization processor, which pitch information corresponds to the average pitch index.
3. The pre-processing system of claim 1 , wherein the pitch equalization processor is configured to equalize by normalizing the speech pitch of the speech signal.
4. The pre-processing system of claim 3 , wherein the pitch equalization processor is configured to normalize the speech pitch of the speech signal to the predefined speech recognition pitch index.
5. The pre-processing system of claim 1 , wherein the pitch equalization processor is configured to filter the speech pitch.
6. The pre-processing system of claim 1 , wherein the pitch equalization processor is configured to provide the pitch-equalized speech signal to one or more of an automatic speech recognition system and an acoustic model system.
7. A method of pre-processing of a speech signal for improved speech recognition, comprising the steps of
receiving the speech signal;
determining a pitch index of the speech signal;
equalizing a speech pitch of the speech signal using pitch information through removal of the speech pitch of the speech signal and re-synthesis of the speech pitch of the speech signal with a predefined speech recognition pitch index; and
providing a pitch-equalized speech signal.
8. The method of claim 7 , further comprising determining the pitch information by calculating an average pitch index from the determined pitch index.
9. A speech processing system, comprising
a pre-processor for improved speech recognition of a speech signal; and
one or more of a speech recognition processor and an acoustic modeler; wherein
the pre-processor comprising at least
a pitch estimation circuit; and
a pitch equalization processor; wherein
the pitch estimation circuit is configured to receive the speech signal to determine a pitch index of the speech signal; and wherein
the pitch equalization processor is configured to receive the speech signal and pitch information;
to equalize a speech pitch of the speech signal using the pitch information through removal of the speech pitch of the speech signal and re-synthesis of the speech pitch of the speech signal with a predefined speech recognition pitch index; and to
provide a pitch-equalized speech signal to one or more of the speech recognition processor and the acoustic modeler.
10. A headset system with a speech processing system of claim 9 .
11. A pre-processing system for improved speech recognition of a speech signal, the system comprising at least
a pitch estimation circuit; and
a pitch classification processor; wherein
the pitch estimation circuit is configured to:
receive the speech signal to determine a pitch index of the speech signal;
to equalize a speech pitch of the speech signal using the pitch information through removal of the speech pitch of the speech signal and re-synthesis of the speech pitch of the speech signal with a predefined speech recognition pitch index; and wherein
the pitch classification processor is configured to
receive pitch information; and to
determine classification information of the speech signal using the pitch information.
12. The pre-processing system of claim 11 , further comprising a pitch averaging circuit; which pitch averaging circuit is configured to receive the pitch index from the pitch estimation circuit, to determine an average pitch index, and to provide the pitch information corresponding to the averaged pitch index to the pitch classification processor.
13. The pre-processing system of claim 11 , wherein the pitch classification processor is configured with multiple pitch bins and is configured to determine the classification information by determining a correlation between one of the pitch bins and the speech signal.
14. The pre-processing system of claim 13 , wherein each pitch bin is associated with one of a plurality of acoustic models of an automatic speech recognition system.
15. The pre-processing system of claim 11 , wherein the pitch classification processor is configured to provide at least the classification information to one or more of an automatic speech recognition system and an acoustic model system.
16. A method of pre-processing of a speech signal for improved speech recognition, comprising the steps of
receiving the speech signal;
determining a pitch index of the speech signal;
equalizing a speech pitch of the speech signal using the pitch information through removal of the speech pitch of the speech signal and re-synthesis of the speech pitch of the speech signal with a predefined speech recognition pitch index; and determining classification information of the speech signal using pitch information.
17. The method of claim 16 , further comprising determining the pitch information by calculating an average pitch index from the determined pitch index.
18. A speech processing system, comprising
a pre-processor for improved speech recognition of a speech signal; and
one or more of a speech recognition processor and an acoustic modeler; wherein
the pre-processor comprising at least
a pitch estimation circuit configured to equalize a speech pitch of the speech signal using the pitch information through removal of the speech pitch of the speech signal and re-synthesis of the speech pitch of the speech signal with a predefined speech recognition pitch index; and
a pitch classification processor; wherein
the pitch estimation circuit is configured to receive the speech signal to determine a pitch index of the speech signal; and wherein
the pitch classification processor is configured to
receive pitch information;
determine classification information of the speech signal using the pitch information; and to
provide the classification information to one or more of the speech recognition processor and the acoustic modeler.
19. A headset system with a speech processing system of claim 18 .
20. The pre-processing system of claim 1 , wherein the pitch index is of a fundamental frequency.
21. The pre-processing system of claim 1 , wherein the pitch estimation circuit is further configured to equalize the speech pitch of the speech signal using the pitch information through application of a fixed pitch filter with the predefined speech recognition pitch index after an average pitch is inverse-filtered.
22. The pre-processing system of claim 1 , wherein the pitch estimation circuit is further configured to equalize the speech pitch to have a fixed pitch.