1. A speech synthesis dictionary creation apparatus for creating speech synthesis dictionaries containing pitch mark data for use in performing speech synthesis by using pitch marks, the apparatus comprising:
first recording means for recording an inter-pitch-mark distance between the first two pitch marks of a voiced portion of speech data to be processed into data for speech synthesis dictionaries;
calculation means for calculating a difference between adjacent inter-pitch-mark distances, which are obtained by calculating distances between adjacent pitch-mark positions; and
second recording means for recording the calculation results obtained by said calculation means in the speech synthesis dictionaries,
wherein the speech synthesis dictionaries are accessed to generate and output synthesized speech.
2. The apparatus according to claim 1, further comprising counting means for counting the number of pitch marks of the voiced portion, and
when the number of pitch marks is counted by said counting means, said second recording means stores the number of pitch marks in a file and manages the number of pitch marks.
3. The apparatus of claim 1, wherein the speech synthesis dictionaries further contain speech data.
4. A method for creating speech synthesis dictionaries containing pitch mark data for use in performing speech synthesis by using pitch marks, the method comprising:
a first recording step for recording an inter-pitch-mark distance between the first two pitch marks of a voiced portion of speech data to be processed into data for speech synthesis dictionaries;
a calculation step for calculating a difference between adjacent inter-pitch-mark distances, which are obtained by calculating distances between adjacent pitch-mark positions; and
a second recording step for recording the calculation results obtained in said calculation step in the speech synthesis dictionaries,
wherein the speech synthesis dictionaries are accessed to generate and output synthesized speech.
5. The method according to claim 4, further comprising a counting step of counting the number of pitch marks of the voiced portion, and
when the number of pitch marks is counted in said counting step, said second recording step stores the number of pitch marks in a file and manages the number of pitch marks.
6. The method of claim 4, wherein the speech synthesis dictionaries further contain speech data.
7. A computer-readable medium storing executable program codes for creating speech synthesis dictionaries, the speech synthesis dictionaries containing pitch mark data for use in performing speech synthesis by using pitch marks, causing a computer to perform the steps comprising:
a first recording step for recording an inter-pitch-mark distance between the first two pitch marks of a voiced portion of speech data to be processed into data for speech synthesis dictionaries;
a calculating step for calculating a difference between adjacent inter-pitch-mark distances, which are obtained by calculating distances between adjacent pitch-mark positions; and
a second recording step for recording the calculation results obtained in said calculating step in the speech synthesis dictionaries.
8. The computer-readable medium of claim 7, wherein the speech synthesis dictionaries further contain speech data.
9. A pitch-mark-data file creation apparatus for creating pitch-mark-data files from speech data, the apparatus comprising:
computer processing means for processing speech data, said computer processing means comprising:
(i) a memory for storing data, including speech data and a pitch-mark-data file, the speech data comprising a voiced portion having pitch marks;
(ii) first determination means for accessing the speech data from said memory and determining inter-pitch-mark distances between adjacent pitch marks of the voiced portion of the speech data;
(iii) first recording means for recording in the pitch-mark-data file a first inter-pitch-mark distance between the first two pitch marks of the voiced portion;
(iv) second determination means for determining a difference between the first inter-pitch-mark distance and a second inter-pitch-mark distance determined by said first determination means; and
(v) second recording means for recording in the pitch-mark-data file the difference determined by said second determination means,
wherein pitch marks of the voiced portion of the speech data can be determined from the first inter-pitch-mark distance determined by said first determination means and the difference determined by said second determination means.
10. The apparatus according to claim 9, further comprising counting means for counting the number of pitch marks of the voiced portion,
wherein said second recording means stores in the pitch-mark-data file the number of pitch marks counted by said counting means.
11. A pitch-mark-data file creation method for an information processing apparatus, the information processing apparatus comprising computer processing means for implementing the method and a memory storing speech data and a pitch-mark-data file, the speech data comprising a voiced portion having pitch marks, the method comprising:
(i) a first determination step of accessing the speech data from the memory and determining inter-pitch-mark distances between adjacent pitch marks of the voiced portion of the speech data;
(ii) a first recording step of recording in the pitch-mark-data file a first inter-pitch-mark distance between the first two pitch marks of the voiced portion;
(iii) a second determination step of determining a difference between the first inter-pitch-mark distance and a second inter-pitch-mark distance determined in said first determination step; and
(iv) a second recording step of recording in the pitch-mark-data file the difference determined in said second determination step,
wherein pitch marks of the voiced portion of the speech data can be determined from the first inter-pitch-mark distance determined in said first determination step and the difference determined in said second determination step.
12. The method according to claim 11, further comprising a counting step of counting the number of pitch marks of the voiced portion,
wherein said recording step further comprises recording in the pitch-mark-data file the number of pitch marks counted in said counting step.
13. A computer-readable medium storing executable program codes for creating pitch-mark-data files for use in performing speech synthesis by using pitch marks, causing a computer to perform the steps comprising:
(i) a first determination step of accessing speech data from a memory and determining inter-pitch-mark distances between adjacent pitch marks of a voiced portion of the speech data;
(ii) a first recording step of recording in a pitch-mark-data file a first inter-pitch-mark distance between the first two pitch marks of the voiced portion;
(iii) a second determination step of determining a difference between the first inter-pitch-mark distance and a second inter-pitch-mark distance determined in said first determination step; and
(iv) a second recording step of recording in the pitch-mark-data file the difference determined in said second determination step,
wherein pitch marks of the voiced portion of the speech data can be determined from the first inter-pitch-mark distance determined in said first determination step and the difference determined in said second determination step.
The claims below are in addition to those above.
All refrences to claim(s) which appear below refer to the numbering after this setence.
1. A method, comprising:
executing, via a processor, a tree traversal operation for a tree data structure, wherein each node in the tree data structure is classified into one of a plurality of nodesets, the plurality of nodesets corresponding to a plurality of local coordinate systems;
receiving, at the processor, a transformation node during the tree traversal operation, wherein the transformation node includes transformation data; and
transforming, by the processor, spatial data included in a query data structure based on the transformation data.
2. The method of claim 1, wherein the tree data structure represents a bounding volume hierarchy (BVH).
3. The method of claim 1, further comprising:
receiving the tree data structure at the processor;
classifying, by the processor, the nodes of the tree data structure into the plurality of nodesets;
selecting, by the processor, a local coordinate system for each nodeset; and
adding at least one transformation node to the tree data structure, each transformation node in the at least one transformation node including transformation data that represents a transformation from one coordinate system into another coordinate system.
4. The method of claim 3, wherein selecting the local coordinate system for each nodeset comprises calculating a geometric center of a set of bounding volumes associated with the nodes in the nodeset, wherein the geometric center is selected as the origin of the local coordinate system.
5. The method of claim 4, wherein selecting the local coordinate system for each nodeset further comprises calculating a scale factor that represents a difference in scale between a global coordinate system and the local coordinate system.
6. The method of claim 3, wherein classifying the nodes of the tree data structure into the plurality of nodesets comprises:
creating an initial proposed classification for the nodes by classifying all nodes in the tree data structure into a single nodeset;
evaluating a cost function for each nodeset in the proposed classification to calculate a set of cost values;
determining that at least one nodeset in the proposed classification did not meet the acceptance criterion; and
adjusting the proposed classification of the nodes into one or more new nodesets, evaluating the cost function for each nodeset in the adjusted proposed classification, and determining whether each of the nodesets in the adjusted proposed classification meets the acceptance criterion.
7. The method of claim 6, wherein adjusting the proposed classification of the nodes comprises, for each nodeset that does not meet the acceptance criterion:
classifying the root node of the nodeset into a top-level nodeset; and
splitting the nodeset into multiple lower-level nodesets, where each nodeset in the multiple lower-level nodesets corresponds to a subtree associated with a child node of the root node.
8. The method of claim 6, wherein the cost function estimates the expected number of additional intersection tests that need to be performed due to conservative rounding.
9. The method of claim 1, wherein transforming spatial data included in the query data structure based on the transformation data comprises transforming the spatial data such that the spatial data is specified relative to a local coordinate system associated with a nodeset connected to the transformation node.
10. The method of claim 1, wherein the processor comprises a parallel processing unit that includes at least one tree traversal unit configured to perform, at least in part, the tree traversal operation.
11. A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform steps comprising:
executing a tree traversal operation for a tree data structure, wherein each node in the tree data structure is classified according to one of a plurality of nodesets, the plurality of nodesets corresponding to a plurality of local coordinate systems;
receiving a transformation node during the tree traversal operation, wherein the transformation node includes transformation data; and
transforming spatial data included in a query data structure based on the transformation data.
12. The non-transitory computer-readable storage medium of claim 11, the steps further comprising:
receiving the tree data structure at the processor;
classifying, by the processor, the nodes of the tree data structure into the plurality of nodesets;
selecting, by the processor, a local coordinate system for each nodeset; and
adding at least one transformation node to the tree data structure, each transformation node in the at least one transformation node including transformation data that represents a transformation from one coordinate system into another coordinate system.
13. The non-transitory computer-readable storage medium of claim 12, wherein selecting the local coordinate system for each nodeset comprises calculating a geometric center of a set of bounding volumes associated with the nodes in the nodeset, wherein the geometric center is selected as the origin of the local coordinate system.
14. The non-transitory computer-readable storage medium of claim 13, wherein selecting the local coordinate system for each nodeset further comprises calculating a scale factor that represents a difference in scale between a global coordinate system and the local coordinate system.
15. The non-transitory computer-readable storage medium of claim 12, wherein classifying the nodes of the tree data structure into the plurality of nodesets comprises:
creating an initial proposed classification for the nodes by classifying all nodes in the tree data structure into a single nodeset;
evaluating a cost function for each nodeset in the proposed classification to calculate a set of cost values;
determining that at least one nodeset in the proposed classification did not meet the acceptance criterion; and
adjusting the proposed classification of the nodes into one or more new nodesets, evaluating the cost function for each nodeset in the adjusted proposed classification, and determining whether each of the nodesets in the adjusted proposed classification meets the acceptance criterion.
16. The method of claim 15, wherein adjusting the proposed classification of the nodes comprises, for each nodeset that does not meet the acceptance criterion:
classifying the root node of the nodeset into a top-level nodeset; and
splitting the nodeset into multiple lower-level nodesets, where each nodeset in the multiple lower-level nodesets corresponds to a subtree associated with a child node of the root node
17. A system, comprising:
a parallel processing unit that includes at least one tree traversal unit configured to:
execute a tree traversal operation for a tree data structure, wherein each node in the tree data structure is classified according to one of a plurality of nodesets, the plurality of nodesets corresponding to a plurality of local coordinate systems,
receive a transformation node during the tree traversal operation, wherein the transformation node includes transformation data, and
transform spatial data included in a query data structure based on the transformation data.
18. The system of claim 17, the parallel processing unit further configured to:
receive the tree data structure;
classify the nodes of the tree data structure into the plurality of nodesets;
select a local coordinate system for each nodeset; and
add at least one transformation node to the tree data structure, each transformation node in the at least one transformation node including transformation data that represents a transformation from one coordinate system into another coordinate system.
19. The system of claim 18, wherein classifying the nodes of the tree data structure into the plurality of nodesets comprises:
creating an initial proposed classification for the nodes by classifying all nodes in the tree data structure into a single nodeset;
evaluating a cost function for each nodeset in the proposed classification to calculate a set of cost values;
determining that at least one nodeset in the proposed classification did not meet the acceptance criterion; and
adjusting the proposed classification of the nodes into one or more new nodesets, evaluating the cost function for each nodeset in the adjusted proposed classification, and determining whether each of the nodesets in the adjusted proposed classification meets the acceptance criterion.
20. The system of claim 17, wherein the tree traversal operation is executed, at least in part, by a tree traversal unit in the parallel processing unit, and wherein a setup unit within the tree traversal unit is configured to transform the spatial data included in the query data structure.