Abstract: A method includes determining a desired power draw for a plurality of loads connected to an electrical grid, each of the plurality of loads connected to the electrical grid through a load supply control and being able to obtain a desired amount of energy from the grid in a desired time period, and transmitting a plurality of instructions through a communication network to a plurality of load supply controls to cause at least some of the loads in the plurality of loads to receive power from the electrical grid at different rates than other loads of the plurality of loads such that the desired power draw is obtained and such that each load of the plurality of loads receives its corresponding desired amount of energy in the desired time period.

Abstract: A method and system is disclosed for simultaneously determining glottal closure instants (GCIs), fundamental frequency (F0s), and voicing state of a speech signal. A speech signal may be processed to determine a sequence of candidate GCIs. For each candidate GCI, a set of candidate F0s may be determined. A lattice of hypotheses may be constructed, where each lattice point is a hypothesis of a concurrence of a candidate GCI, a candidate F0, and voicing state. Each given hypothesis may also include a score of the candidate GCI, F0, and voicing state for evaluating a cost of the given hypothesis and a cost of connections between the given hypothesis and other hypotheses of the lattice. Dynamic programming may be used to determine a least-cost path through the lattice, and backtracking across the path may be used to determine an optimal set of GCIs, F0s and voicing states of the speech signal.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for performing audio subsystem enhancement. In one aspect, a method includes: receiving a voice search query by an automatic speech recognition (ASR) engine that processes voice search queries for a search engine, wherein the voice search query includes an audio signal that corresponds to an utterance, and a test flag that indicates that an audio test is being performed; performing speech recognition on the audio signal to select one or more textual, candidate transcriptions that match the utterance; generating, in response to receiving the test flag, one or more audio quality metrics using the audio signal; and generating a response to the voice search query by the ASR engine, wherein the response references one or more of the candidate transcriptions and one or more of the audio quality metrics.

Abstract: A method includes determining a desired power draw for a plurality of loads connected to an electrical grid, each of the plurality of loads connected to the electrical grid through a load supply control and being able to obtain a desired amount of energy from the grid in a desired time period, and transmitting a plurality of instructions through a communication network to a plurality of load supply controls to cause at least some of the loads in the plurality of loads to receive power from the electrical grid at different rates than other loads of the plurality of loads such that the desired power draw is obtained and such that each load of the plurality of loads receives its corresponding desired amount of energy in the desired time period.

Abstract: A method includes determining a desired power draw for a plurality of loads connected to an electrical grid, each of the plurality of loads connected to the electrical grid through a load supply control and being able to obtain a desired amount of energy from the grid in a desired time period, and transmitting a plurality of instructions through a communication network to a plurality of load supply controls to cause at least some of the loads in the plurality of loads to receive power from the electrical grid at different rates than other loads of the plurality of loads such that the desired power draw is obtained and such that each load of the plurality of loads receives its corresponding desired amount of energy in the desired time period.

Abstract: A method includes determining a desired power draw for a plurality of loads connected to an electrical grid, each of the plurality of loads connected to the electrical grid through a load supply control and being able to obtain a desired amount of energy from the grid in a desired time period, and transmitting a plurality of instructions through a communication network to a plurality of load supply controls to cause at least some of the loads in the plurality of loads to receive power from the electrical grid at different rates than other loads of the plurality of loads such that the desired power draw is obtained and such that each load of the plurality of loads receives its corresponding desired amount of energy in the desired time period.

Abstract: A method of smoothing fundamental frequency discontinuities at boundaries of concatenated speech segments includes determining, for each speech segment, a beginning fundamental frequency value and an ending fundamental frequency value. The method further includes adjusting the fundamental frequency contour of each of the speech segments according to a linear function calculated for each particular speech segment, and dependent on the beginning and ending fundamental frequency values of the corresponding speech segment. The method calculates the linear function for each speech segment according to a coupled spring model with three springs for each segment. A first spring constant, associated with the first spring and the second spring, is proportional to a duration of voicing in the associated speech segment. A second spring constant, associated with the third spring, models a non-linear restoring force that resists a change in slope of the segment fundamental frequency contour.

Abstract: A method of smoothing fundamental frequency discontinuities at boundaries of concatenated speech segments includes determining, for each speech segment, a beginning fundamental frequency value and an ending fundamental frequency value. The method further includes adjusting the fundamental frequency contour of each of the speech segments according to a linear function calculated for each particular speech segment, and dependent on the beginning and ending fundamental frequency values of the corresponding speech segment. The method calculates the linear function for each speech segment according to a coupled spring model with three springs for each segment. A first spring constant, associated with the first spring and the second spring, is proportional to a duration of voicing in the associated speech segment. A second spring constant, associated with the third spring, models a non-linear restoring force that resists a change in slope of the segment fundamental frequency contour.