1. A method comprising:
approximating a posterior probability for a switching state space model, the posterior probability providing the likelihood of a set of speech units and a set of hidden parameters for a sequence of frames of speech based upon input values associated with the sequence of frames of speech, wherein approximating the posterior probability comprises multiplying individual hidden parameter probabilities together to form a product of the hidden parameter probabilities, wherein each individual hidden parameter probability provides the probability of a hidden parameter for a frame given a speech unit of the frame and given the input values for the sequence of frames;
adjusting parameters that define the hidden parameter probabilities so that the hidden parameter probabilities provide a second approximation of the posterior probability; and
performing speech recognition to output a sequence of words represented by the frames of speech by using the second approximation of the posterior probability to produce a sequence of speech units represented by the frames of speech.
2. The method of claim 1 wherein adjusting the parameters that define the hidden parameter probabilities comprises adjusting the parameters based on parameters of the switching state space model to produce adjusted parameters.
3. The method of claim 2 wherein the input values are generated from a framing speech signal and wherein adjusting the parameters further comprises adjusting the parameters of the switching state space model based on adjusted parameters that define the hidden parameter probabilities to form adjusted switching state space model parameters and adjusting the adjusted parameters that define the hidden parameter probabilities based on the adjusted switching state space model parameters.
4. The method of claim 1 further comprising using the approximation of the posterior probability to identify a sequence of hidden parameters for the sequence of frames.
5. A computer-readable storage medium having encoded thereon computer-executable instructions that when executed by a processor cause the processor to perform steps comprising:
defining a hidden dynamic model of speech comprising a model of a hidden production-related parameter;
approximating a posterior probability that provides the likelihood of a sequence of the hidden production-related parameters and a sequence of speech units based on a sequence of input values of a speech signal without fixing the boundaries of the speech units by approximating the posterior probability using a product of hidden production-related parameter probabilities, each hidden production-related parameter probability providing the likelihood of a hidden production-related parameter for a frame of the input values given a speech unit for the frame and the sequence of input values;
determining model parameters that describe the hidden production-related parameter probabilities in part by determining a speech unit probability that describes the likelihood of a speech unit for a frame given the sequence of input values for each of a plurality of speech units for each frame of the speech signal; and
performing speech recognition to output a sequence of words represented by the speech signal by selecting a speech unit for each frame of the speech signal based at least in part on the speech unit probabilities.
6. The computer-readable storage medium of claim 5 wherein using the speech unit probability to select a speech unit comprises selecting the speech unit with the highest speech unit probability at each frame.
7. The computer-readable storage medium of claim 5 further comprising applying a duration constraint to the speech units and wherein using the speech unit probability to select a speech unit comprises selecting a best sequence of speech units based on the duration constraint and the plurality of speech unit probabilities for each frame.
8. The computer-readable storage medium of claim 5 wherein approximating the posterior probability further comprises determining a speech unit transition probability that describes the likelihood of a speech unit for a frame given a speech unit of another frame and the sequence of input values.
9. The computer-readable storage medium of claim 8 wherein selecting a speech unit for each frame further comprises using the speech unit transition probability to select a speech unit for a frame.
10. The computer-readable storage medium of claim 5 wherein determining model parameters that describe the likelihood of a hidden production-related parameter comprises iteratively adjusting the model parameters based on model parameters for the hidden dynamic model of speech and previous values of the model parameters that describe the likelihood of a production-related parameter.
11. The computer-readable storage medium of claim 10 further comprising adjusting the model parameters for the hidden dynamic model of speech based on the parameters that describe the likelihood of a hidden production-related parameter.
12. The computer-readable storage medium of claim 5 further comprising identifying a hidden production-related parameter for each frame of the input values based on the model parameters that describe the likelihood of a hidden production-related parameter.
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 for cooling an object with a cryogen having a critical point defined by a critical-point pressure and a critical-point temperature, the method comprising:
raising a pressure of the cryogen to between 0.8 and 1.2 of its critical-point pressure and providing the cryogen at a temperature within \xb110% of its critical-point temperature;
thereafter, reducing a temperature of the cryogen without decreasing the pressure of the cryogen below 0.8 times its critical-point pressure;
thereafter, placing the cryogen in thermal communication with the object to increase the temperature of the cryogen to an ambient temperature along a thermodynamic path that maintains a pressure greater than 0.8 times the critical-point pressure for a duration that the cryogen and object are in thermal communication;
thereafter, removing the cryogen from thermal communication with the object; and
thereafter, reducing the pressure of the cryogen to an ambient pressure.
2. The method recited in claim 1 wherein reducing the temperature of the cryogen comprises placing the cryogen in thermal communication with liquid cryogen having the same chemical structure maintained at substantially the ambient pressure.
3. The method recited in claim 2 wherein the determined pressure value is between 0.8 and 1.2 times the critical-point pressure.
4. The method recited in claim 2 wherein raising the pressure of the cryogen comprises providing the cryogen at a temperature within \xb110% of its critical-point temperature.
5. The method recited in claim 1 wherein the cryogen is selected from the group consisting of N2, SF6, N2O, He, and CO2.
6. The method recited in claim 1 wherein:
raising the pressure of the cryogen comprises raising the pressure of the cryogen to greater than a pressure value determined to provide the cryogen at a reduced molar volume that prevents vapor lock; and
the thermodynamic path maintains the pressure above the determined pressure value.
7. The method recited in claim 1 wherein raising the pressure of the cryogen comprises:
placing the cryogen within a thermally insulated tank; and
applying heat within the thermally insulated tank at least until a predetermined pressure within the thermally insulated tank is reached.
8. The method recited in claim 1 wherein the cryogen comprises N2.
9. The method recited in claim 1 wherein the pressure of the cryogen is substantially constant for the duration that the cryogen and object are in thermal communication.
10. The method recited in claim 1 wherein reducing the temperature of the cryogen comprises placing the cryogen in thermal communication with a second liquid cryogen having a temperature lower than the temperature of the cryogen.
11. The method recited in claim 10 wherein the cryogen and second cryogen are chemically identical.
12. The method recited in claim 10 wherein the second liquid cryogen is substantially at ambient pressure.
13. The method recited in claim 10 wherein the pressure of the cryogen is substantially constant while reducing the temperature of the cryogen.