1. A method for recognizing speech, the method comprising:
recognizing a sequence of words;
processing the sequence of words using word agglomeration that replaces the sequence of words with an associated n-tuple sequence, the n-tuple sequence comprising word n-tuples that are all strings of n consecutive words in the sequence of words, wherein the n-tuple sequence is represented by a vector representation in a semantic space; and
classifying the processed sequence of words as a predetermined command based on the vector representation of the n-tuple sequence in the semantic space.
2. The method of claim 1, further comprising performing an action corresponding to the predetermined command.
3. The method of claim 1, wherein classifying comprises semantically inferring the predetermined command from the associated word n-tuple sequence.
4. The method of claim 1, wherein classifying comprises semantically inferring the predetermined command from the processed sequence of words.
5. The method of claim 4, wherein semantically inferring comprises determining a correlation between the processed sequence of words and at least one semantic anchor.
6. The method of claim 5, wherein the correlation is a distance between the vector corresponding to the processed sequence of words and a vector corresponding to the at least one semantic anchor.
7. The method of claim 6, wherein semantically inferring further comprises selecting the predetermined command from the semantic anchor vector having the shortest distance.
8. The method of claim 7, wherein the semantic anchor represents a one of a plurality of predetermined commands.
9. The method of claim 5, wherein the at least one semantic anchor is derived from a training data.
10. The method of claim 4, wherein semantically inferring the predetermined command depends on the order of the words in the processed sequence of words.
11. The method of claim 1, wherein the classifying comprises:
generating the vector representation of the processed sequence of words; and
comparing the vector representation to a plurality of semantic anchors, wherein each of the plurality of semantic anchors corresponds to one of a plurality of voice commands.
12. The method of claim 11, wherein the classifying further comprises:
selecting a semantic anchor from the plurality of semantic anchors that is most similar to the vector representation; and
classifying the processed sequence of words as the command that corresponds to the selected semantic anchor.
13. The method of claim 12, wherein the selecting comprises:
for each of the plurality of semantic anchors, identifying the similarity between the vector representation and the semantic anchor by calculating the cosine of the angle between the product of the vector representation and a diagonal matrix of singular values and the product of the semantic anchor and the diagonal matrix of singular values; and
selecting the semantic anchor from the plurality of semantic anchors that corresponds to the largest cosine value as the semantic anchor that is most similar to the vector representation.
14. The method of claim 11, wherein the vector representation is an indication of how frequently each of a plurality of word n-tuples occurs within the processed sequence of words.
15. The method of claim 14, wherein each of the plurality of semantic anchors is an indication of how frequently each of the plurality of word n-tuples occurs with respect to the corresponding command.
16. The method of claim 11, wherein each of the plurality of semantic anchors represents a plurality of different ways of speaking the corresponding command.
17. The method of claim 11, wherein each of the plurality of semantic anchors represents a plurality of different commands having the same words, but in a different order.
18. A machine-readable medium having stored thereon a plurality of instructions that, when executed by a processor, cause the processor to recognize a voice command by:
recognizing a sequence of words;
processing the sequence of words using word agglomeration that replaces the sequence of words with an associated n-tuple sequence, the n-tuple sequence comprising word n-tuples that are all strings of n consecutive words in the sequence of words, wherein the n-tuple sequence is represented by a vector representation in a semantic space; and
classifying the processed sequence of words as a predetermined command based on the vector representation of the n-tuple sequence in the semantic space.
19. The machine-readable medium of claim 18, wherein the plurality of instructions further cause the processor to perform an action corresponding to the predetermined command.
20. The machine-readable medium of claim 18, wherein the instructions for classifying comprises semantically inferring the predetermined command from the associated word n-tuple sequence.
21. The machine-readable medium of claim 18, wherein the instructions for semantically inferring comprises determining a correlation between a semantic representation of the associated word n-tuple sequence and at least one semantic anchor.
22. The machine-readable medium of claim 21, wherein the instructions for determining a correlation comprise determining a distance between the vector corresponding to the semantic representation and a vector corresponding to the at least one semantic anchor.
23. The machine-readable medium of claim 22, wherein the instructions for semantically inferring further comprises selecting the predetermined command from the semantic anchor vector having the shortest distance to the vector corresponding to the semantic representation.
24. The machine-readable medium of claim 21, wherein the at least one semantic anchor represents a one of a plurality of predetermined commands.
25. The machine-readable medium of claim 21, wherein the at least one semantic anchor is derived from a training data.
26. The machine-readable medium of claim 21, wherein the performance of the instructions for semantically inferring the predetermined command depends on the order of the words in the processed sequence of words.
27. An apparatus for recognizing a voice command, the apparatus comprising:
a speech recognizer to recognize a sequence of words received as the voice command;
a processor to process the sequence of words using word agglomeration that replaces the sequence of words with an n-tuple sequence, the n-tuple sequence comprising word n-tuples that are all strings of n consecutive words in the sequence of words, wherein the n-tuple sequence is represented by a vector representation in a semantic space; and
a semantic classifier, coupled to the processor, to semantically infer from the representation of the n-tuple sequence in the semantic space which of a plurality of predetermined commands correlate to the voice command.
28. The apparatus of claim 27, further comprising:
an action generator, coupled to the semantic classifier, to use the vector representation to determine an action to be performed.
29. The apparatus of claim 27, wherein the semantic classifier is further to compare the vector representation to a plurality of semantic anchors, wherein each of the plurality of semantic anchors corresponds to a one of the plurality of predetermined commands.
30. The apparatus of claim 29, wherein the semantic classifier is further to identify a semantic anchor from the plurality of semantic anchors that is most similar to the vector representation, and to classify the vector representation as the one of the plurality of predetermined commands that corresponds to the identified semantic anchor.
31. An apparatus for recognizing a voice command, the apparatus comprising:
means for recognizing a sequence of words received as the voice command;
means for processing the sequence of words using word agglomeration that replaces the sequence of words with an n-tuple sequence, the n-tuple sequence comprising word n-tuples that are all strings of n consecutive words in the sequence of words, wherein the n-tuple sequence is represented by a vector representation
means, coupled to the means for processing, for semantically inferring from the vector representation of the n-tuple sequence in the semantic space which of a plurality of predetermined commands correlate to the voice command.
32. The apparatus of claim 31, further comprising means, coupled to the means for semantically inferring, for using the vector representation to determine an action to be performed.
33. The apparatus of claim 31, wherein the means for semantically inferring further comprises:
means for comparing the vector representation to a plurality of semantic anchors, wherein each of the plurality of semantic anchors corresponds to one of a plurality of predetermined commands.
34. The apparatus of claim 33, wherein the means for semantically inferring further comprises:
means for identifying a semantic anchor of the plurality of semantic anchors that is most similar to the vector representation; and
means for classifying the vector representation as the one of the plurality of predetermined commands that corresponds to the identified semantic anchor.
The claims below are in addition to those above.
All refrences to claim(s) which appear below refer to the numbering after this setence.
What is claimed is:
1. An elastically stretchable composite sheet comprising an elastic layer, said elastic layer being elastically stretchable at least in one direction, and an inelastic layer of fibers, said inelastic layer being inelastically stretchable in said one direction and bonded to least on one surface of said elastic layer, and at least 80% by weight of said fiber comprising thermoplastic synthetic fiber which contains lubricant of 0.15.0% by weight.
2. The composite sheet according to claim 1, wherein said thermoplastic synthetic fiber is selected from a group including fiber made of polypropylene’s homopolymer, ethylenepropylene copolymer, ethylenepropylenebutene copolymer and fibrous mixture comprising at least two of them.
3. The composite sheet according to claim 1, wherein said elastic layer and said inelastic layer are fused together in bond regions intermittently arranged in said one direction and a length of said fiber extending between each pair of the adjacent bond regions in said one direction is longer than a rectilinear distance between said pair of the adjacent bond regions.
4. The composite sheet according to claim 1, wherein said lubricant is selected from a group including fatty acid amide, fatty acid ester and metallic soap.
5. A process for making an elastically stretchable composite sheet comprising the steps of feeding an elastic web which is elastically stretchable at least in one direction, placing an inelastically stretchable fibrous layer upon at least one surface of said elastic web and intermittently bonding these both webs to each other in said one direction, said process further comprising:
at least 80% by weight of said fiber comprises thermoplastic synthetic fiber containing lubricant of 0.15.0% by weight.
6. The process according to claim 5, wherein said thermoplastic synthetic fiber is selected from a group including fiber made of polypropylene’s homopolymer, ethylenepropylene copolymer, ethylenepropylenebutene copolymer and fibrous mixture comprising at least two of them.
7. The process according to claim 5, further including the steps of stretching said elastic web and said inelastic web bonded together at least by 80% in said one direction and then leaving said webs to contract under a contractile force of said elastic layer.