Techniques for enhancing the performance of concatenative speech synthesis

• US 20040024600A1
• Filed: 07/30/2002
• Published: 02/05/2004
• Est. Priority Date: 07/30/2002
• Status: Active Grant

• Current Assignee: Cerence Operating Company (Cerence Inc.)
• Original Assignee: International Business Machines Corporation
• Inventors: Hamza, Wael Mohamed, Picheny, Michael Alan
• Granted Patent: US 8,145,491 B2
• Time in Patent Office: Days
• US Class Current: 704/268
• CPC Class Codes: G10L 13/07 Concatenation rules

Abstract

When pitch of a speech segment is being modified from a current pitch to a requested pitch, and the difference between these is relatively large, a pitch modification algorithm is used to modify the pitch of the speech segment. When the difference between current and requested pitches is relatively small, the pitch of the speech segment is not modified. After one or the other speech modification techniques are used, then the resultant modified speech segment is overlapped and added to previously modified speech segments. A modification ratio is determined in order to quantify the difference between the current and requested pitches for a speech segment. The modification ratio is a ratio between the requested and current pitches. Low and high ratio thresholds are used to determine when pitch is being modified to a predetermined high degree, and whether pitch of the speech segment will or will not be modified.

Claims

1. A method for use with speech synthesis, comprising the steps of:

determining a value indicating how much pitch is to be modified for a current speech segment; and

selecting one of a plurality of speech segment modification techniques based on the value.

2. The method of claim 1, wherein each of the speech segment modification techniques produces a modified speech segment suitable for use with a subsequent overlap and add step.

3. The method of claim 1, wherein the step of determining a value further comprises the steps of:

determining an original pitch value; and

determining a requested pitch value.

4. The method of claim 3, wherein the step of determining an original pitch value comprises the step of subtracting a next pitch mark from a current pitch mark to determine the original pitch value.

5. The method of claim 3, wherein the step of determining a requested pitch value further comprises extracting the requested pitch value from a requested pitch contour.

6. The method of claim 1, wherein the value is a pitch ratio.

7. The method of claim 6, wherein the pitch ratio is determined by dividing a requested pitch by a current pitch.

8. The method of claim 1, wherein the step of selecting further comprises the steps of:

selecting a first speech segment modification technique when the value is within a predetermined range; and

selecting a second speech segment modification technique when the value is not within the predetermined range.

9. The method of claim 8, wherein the first speech segment modification technique comprises the step of windowing the current speech segment with a window function.

10. The method of claim 8, wherein the second speech segment modification technique comprises the steps of windowing the current speech segment with a window function and modifying the pitch of the windowed current speech segment.

11. The method of claim 8, wherein the predetermined range is between high and low ratio thresholds.

12. The method of claim 11, wherein the high and low ratio thresholds are determined experimentally.

13. The method of claim 11, wherein the high and low ratio thresholds are determined for speech segments from a particular voice or voices.

14. The method of claim 8, wherein each of the first and second speech segment modification techniques produce a modified speech segment, and wherein the method further comprises the step of overlapping and adding the modified speech segment to a previously modified speech segment.

15. The method of claim 14, wherein the overlap and add of the modified speech segment is performed at a synthesis time instant.

16. The method of claim 15, further comprising the step of incrementing the synthesis time instant by an increment, wherein the increment is set for the first speech segment modification technique via a subtraction between a next pitch mark and a present pitch mark and wherein the increment for the second speech modification technique is determined by a pitch modification algorithm.

17. The method of claim 8, wherein the first speech segment modification technique comprises a pitch modification algorithm.

18. The method of claim 8, wherein the second speech segment modification technique comprises a pitch modification algorithm.

19. An apparatus for use with speech synthesis, comprising:

at least one processor operable to;

determine a value indicating how much pitch is to be modified for a current speech segment; and

select one of a plurality of speech segment modification techniques based on the value.

20. The apparatus of claim 19, wherein each of the speech segment modification techniques produces a modified speech segment suitable for use with an overlap and add module.

21. The apparatus of claim 19, wherein the at least one processor is further operable, when determining a value, to:

determine an original pitch value; and

determine a requested pitch value.

22. The apparatus of claim 19, wherein the value is a pitch ratio.

23. The apparatus of claim 22, wherein the pitch ratio is determined by dividing a requested pitch by a current pitch.

24. The apparatus of claim 19, wherein the at least one processor is further operable, when selecting, to:

select a first speech segment modification technique when the value is within a predetermined range; and

select a second speech segment modification technique when the value is not within the predetermined range.

25. An article of manufacture for use with speech synthesis, comprising:

a computer-readable medium having computer-readable code means embodied thereon, the computer-readable program code means comprising;

a step to determine a value indicating how much pitch is to be modified for a current speech segment; and

a step to select one of a plurality of speech segment modification techniques based on the value.

26. The article of claim 25, wherein each of the speech segment modification techniques produces a modified speech segment suitable for use with a subsequent overlap and add step.

27. The article of claim 25, wherein the computer-readable program code means, when determining a value, further comprises:

a step to determine an original pitch value; and

a step to determine a requested pitch value.

28. The article of claim 25, wherein the value is a pitch ratio.

29. The article of claim 28, the computer-readable program code means further comprises a step to determine the pitch ratio by dividing a requested pitch by a current pitch.

30. The article of claim 25, wherein the computer-readable program code means, when selecting, further comprises:

a step to select a first speech segment modification technique when the value is within a predetermined range; and

a step to select a second speech segment modification technique when the value is not within the predetermined range.