Abstract: A method using drive test data for propagation model calibration includes: step 1, obtaining original drive test data; step 2, selecting the data from the drive test data according to predefined conditions as effective drive test data; and step 3, extracting the effective drive test data to form a data file used for propagation model calibration. An apparatus using drive test data for propagation model calibration includes: a drive test data obtaining module, configured to obtain the drive test data in the regions to be calibrated; an effective drive test data generation module, configured to generate effective drive test data from the drive test data according to predefined conditions; and a data file generation module, configured to extract the effective drive test data to form a data file used for propagation model calibration. The present invention utilizes drive test data of existing networks to largely decrease the CW test work and reduce the network building cost.

Abstract: Methods of forming a field-effect transistor and structures for a field-effect transistor. A gate structure is formed that overlaps with a channel region beneath a top surface of a semiconductor fin. The semiconductor fin is etched with an anisotropic etching process to form a cavity having a sidewall with a planar section extending vertically toward the top surface of the semiconductor fin and adjacent to the channel region in the semiconductor fin. The semiconductor fin is then etched with an isotropic etching process that widens the cavity at the top surface while preserving verticality of the planar section.

Abstract: One illustrative device disclosed herein includes, among other things, a fin defined on a substrate. A gate electrode structure is positioned above the fin in a channel region. A source/drain region is defined in the fin. The source/drain region includes a first epitaxial semiconductor material. The first epitaxial semiconductor material includes a dopant species having a first concentration. A diffusion blocking layer is positioned above the first epitaxial semiconductor material. A second epitaxial semiconductor material is positioned above the diffusion blocking layer. The second epitaxial semiconductor material includes the dopant species having a second concentration greater than the first concentration.

Abstract: The technology relates to converting text to speech utilizing recurrent neural networks (RNNs). The recurrent neural networks may be implemented as multiple modules for determining properties of the text. In embodiments, a part-of-speech RNN module, letter-to-sound RNN module, a linguistic prosody tagger RNN module, and a context awareness and semantic mining RNN module may all be utilized. The properties from the RNN modules are processed by a hyper-structure RNN module that determine the phonetic properties of the input text based on the outputs of the other RNN modules. The hyper-structure RNN module may generate a generation sequence that is capable of being converting to audible speech by a speech synthesizer. The generation sequence may also be optimized by a global optimization module prior to being synthesized into audible speech.

Abstract: An example text-to-speech learning system performs a method for generating a pronunciation sequence conversion model. The method includes generating a first pronunciation sequence from a speech input of a training pair and generating a second pronunciation sequence from a text input of the training pair. The method also includes determining a pronunciation sequence difference between the first pronunciation sequence and the second pronunciation sequence; and generating a pronunciation sequence conversion model based on the pronunciation sequence difference. An example speech recognition learning system performs a method for generating a pronunciation sequence conversion model. The method includes extracting an audio signal vector from a speech input and applying an audio signal conversion model to the audio signal vector to generate a converted audio signal vector. The method also includes adapting an acoustic model based on the converted audio signal vector to generate an adapted acoustic model.

Abstract: Methods to reduce a width of a channel region of Si fins and the resulting devices are disclosed. Embodiments include forming a Si fin in a Si layer; forming a channel region over the Si fin including a dummy gate with a spacer on each side; forming S/D regions at opposite ends of the Si fin; removing the dummy gate, forming a cavity; thinning sidewalls of the Si fin; and forming a high-k/metal gate in the cavity.

Abstract: One illustrative device disclosed herein includes, among other things, a fin defined on a substrate. A gate electrode structure is positioned above the fin in a channel region. A source/drain region is defined in the fin. The source/drain region includes a first epitaxial semiconductor material. The first epitaxial semiconductor material includes a dopant species having a first concentration. A diffusion blocking layer is positioned above the first epitaxial semiconductor material. A second epitaxial semiconductor material is positioned above the diffusion blocking layer. The second epitaxial semiconductor material includes the dopant species having a second concentration greater than the first concentration.

Abstract: A method includes forming a gate electrode structure above a channel region defined in a semiconductor material. The semiconductor material is recessed in a source/drain region. A first material is epitaxially grown in the source/drain region. The first material includes a dopant species having a first concentration. A diffusion blocking layer is formed in the source/drain region above the first material. A second material is epitaxially grown in the source/drain region above the diffusion blocking layer. The second material comprises the dopant species having a second concentration greater than the first concentration.

Abstract: A terminal device, system and method for accessing a virtual desktop, the terminal device includes: a plurality of peripheral interface groups, configured to connect with a plurality of peripheral groups, wherein each of the peripheral groups comprises one or more input apparatus; a plurality of virtual machines, configured to control corresponding desktops supplied by a virtual desktop server system according to operating data input by the peripheral groups; and a network interface, configured to connect with the virtual desktop server system. With the above terminal device, system and method, the cost of the current virtual desktop system can be reduced effectively.

Abstract: An example text-to-speech learning system performs a method for generating a pronunciation sequence conversion model. The method includes generating a first pronunciation sequence from a speech input of a training pair and generating a second pronunciation sequence from a text input of the training pair. The method also includes determining a pronunciation sequence difference between the first pronunciation sequence and the second pronunciation sequence; and generating a pronunciation sequence conversion model based on the pronunciation sequence difference. An example speech recognition learning system performs a method for generating a pronunciation sequence conversion model. The method includes extracting an audio signal vector from a speech input and applying an audio signal conversion model to the audio signal vector to generate a converted audio signal vector. The method also includes adapting an acoustic model based on the converted audio signal vector to generate an adapted acoustic model.

Abstract: Methods to reduce a width of a channel region of Si fins and the resulting devices are disclosed. Embodiments include forming a Si fin in a Si layer; forming a channel region over the Si fin including a dummy gate with a spacer on each side; forming S/D regions at opposite ends of the Si fin; removing the dummy gate, forming a cavity; thinning sidewalls of the Si fin; and forming a high-k/metal gate in the cavity.

Abstract: A splicing method and splicing system are described for HTTP live streaming media fragmentation.

Abstract: A method using drive test data for propagation model calibration includes: step 1, obtaining original drive test data; step 2, selecting the data from the drive test data according to predefined conditions as effective drive test data; and step 3, extracting the effective drive test data to form a data file used for propagation model calibration. An apparatus using drive test data for propagation model calibration includes: a drive test data obtaining module, configured to obtain the drive test data in the regions to be calibrated; an effective drive test data generation module, configured to generate effective drive test data from the drive test data according to predefined conditions; and a data file generation module, configured to extract the effective drive test data to form a data file used for propagation model calibration. The present invention utilizes drive test data of existing networks to largely decrease the CW test work and reduce the network building cost.

Abstract: A method includes forming a gate electrode structure above a channel region defined in a semiconductor material. The semiconductor material is recessed in a source/drain region. A first material is epitaxially grown in the source/drain region. The first material includes a dopant species having a first concentration. A diffusion blocking layer is formed in the source/drain region above the first material. A second material is epitaxially grown in the source/drain region above the diffusion blocking layer. The second material comprises the dopant species having a second concentration greater than the first concentration.

Abstract: Pronunciation issues for synthesized speech are automatically detected using human recordings as a reference within a Speech Recognition Assisted Evaluation (SRAE) framework including a Text-To-Speech flow and a Speech Recognition (SR) flow. A pronunciation issue detector evaluates results obtained at multiple levels of the TTS flow and the SR flow (e.g. phone, word, and signal level) by using the corresponding human recordings as the reference for the synthesized speech, and outputs possible pronunciation issues. A signal level may be used to determine similarities/differences between the recordings and the TTS output. A model level checker may provide results to the pronunciation issue detector to check the similarities of the TTS and the SR phone set including mapping relations. Results from a comparison of the SR output and the recordings may also be evaluation by the pronunciation issue detector. The pronunciation issue detector outputs a list that lists potential pronunciation issue candidates.

Abstract: The technology relates to converting text to speech utilizing recurrent neural networks (RNNs). The recurrent neural networks may be implemented as multiple modules for determining properties of the text. In embodiments, a part-of-speech RNN module, letter-to-sound RNN module, a linguistic prosody tagger RNN module, and a context awareness and semantic mining RNN module may all be utilized. The properties from the RNN modules are processed by a hyper-structure RNN module that determine the phonetic properties of the input text based on the outputs of the other RNN modules. The hyper-structure RNN module may generate a generation sequence that is capable of being converting to audible speech by a speech synthesizer. The generation sequence may also be optimized by a global optimization module prior to being synthesized into audible speech.

Abstract: The technology relates to performing letter-to-sound conversion utilizing recurrent neural networks (RNNs). The RNNs may be implemented as RNN modules for letter-to-sound conversion. The RNN modules receive text input and convert the text to corresponding phonemes. In determining the corresponding phonemes, the RNN modules may analyze the letters of the text and the letters surrounding the text being analyzed. The RNN modules may also analyze the letters of the text in reverse order. The RNN modules may also receive contextual information about the input text. The letter-to-sound conversion may then also be based on the contextual information that is received. The determined phonemes may be utilized to generate synthesized speech from the input text.

Abstract: A terminal device, system and method for accessing a virtual desktop, the terminal device includes: a plurality of peripheral interface groups, configured to connect with a plurality of peripheral groups, wherein each of the peripheral groups comprises one or more input apparatus; a plurality of virtual machines, configured to control corresponding desktops supplied by a virtual desktop server system according to operating data input by the peripheral groups; and a network interface, configured to connect with the virtual desktop server system. With the above terminal device, system and method, the cost of the current virtual desktop system can be reduced effectively.

Abstract: The invention belongs to the technical field of rapid solidification of amorphous alloy and concretely relates to an iron-based amorphous alloy broad ribbon, wherein the width is 220-1000 mm, the thickness is 0.02-0.03 mm, the transversal thickness deviation is smaller than +/?0.002 mm, the lamination factor is larger than 0.84, the saturation magnetic-flux density is larger than 1.5 T, the iron loss is smaller than 0.20 W/kg under the conditions that the frequency is 50 Hz and the maximum magnetic-flux density is 1.3 T, and the exciting power is smaller than 0.50 VA/kg. The invention also relates to a manufacturing method of the broad ribbon, and a single-roll quenching method is adopted, wherein the width of a nozzle slot is 0.4-0.7 mm, the transversal width deviation of the nozzle slot is smaller than +/?0.05 mm, the transversal flatness deviation of a cooling roll (4) is smaller than 0.02 mm, and the surface roughness Ra is smaller than 0.0005 mm.

Abstract: A method includes performing blind spectrum sensing of a frequency band to determine whether a primary user is using the frequency band. The blind spectrum sensing is based at least in part on a comparison between a detection statistic based on a maximum correlation coefficient and a detection threshold based on theoretical computation of a distribution of the detection statistic. The maximum correlation coefficient is for correlations between a plurality of signals corresponding to a plurality of snapshots taken by a cognitive radio of the frequency band and corresponding to a plurality of antennas used by the cognitive radio for taking the snapshots. The method includes determining whether to communicate using the frequency band based on whether the blind spectrum sensing indicates the frequency band is or is not used by the primary user. Apparatus and program products are disclosed.