Future Wei Technologies, Inc.

Huawei Technologies Co., Limited

Ping an Technology (Shenzhen) Co., Ltd.

Zte Corporation

Ning Cheng

Frank J Effenberger

Xiang Liu

Jing Jing Xiao

Jianzong Wang

Zhenxing Liao

Zhenping Wang

Hao Liang

Lei Zhou

Huafeng Lin

Guo Ji Wei

Guikai Peng

Yuanqiu Luo

Huaiyu Zeng

Xiaofeng Zhang

Fei Ye

Bin Guo

Jin Yang

Min Zhou

Lei Jing

Yinjun Han

Yongxi Tan

Andy Shen

Meng Sui

Naresh Chand

Dianbo Zhao

Rujun Li

Haipeng Li

Hong Gao

Xianli Hu

Zhenghua Chen

Peter Ossieur

Zhishan Feng

Yuanzhe Cai

Dajiang Tang

A semiconductor laser comprises: a substrate; a first cladding layer disposed above the substrate; a second cladding layer disposed above the first cladding layer so that the first cladding layer is positioned between the substrate and the second cladding layer; and a first mode expansion layer within the first cladding layer, a second mode expansion layer within the second cladding layer, or both the first mode expansion layer within the first cladding layer and the second mode expansion layer within the second cladding.

Semiconductor laser with a mode expansion layer

An apparatus comprises a first mirror; a second mirror; a modulation layer positioned between the first mirror and the second mirror and comprising a plurality of modulation regions; a diffraction layer positioned between the modulation layer and the second mirror, and an input port admitting a light beam into the apparatus. The light beam passes through the diffraction layer and is modulated by the modulation layer to create a first modulated beam before being reflected by the first mirror, the first mirror reflecting the first modulated beam toward the second mirror, the second mirror reflecting the first modulated beam toward the modulation layer to be modulated for at least a second time.

Diffractive deep neural network (D2NN) processing using a single modulation layer

An OLT comprises a processor configured to: obtain optical powers associated with ONUs, and generate an instruction instructing a transmit order of transmissions based on the optical powers; a transmitter coupled to the processor and configured to transmit the instruction to the ONUs; and a receiver coupled to the processor and configured to receive the transmissions from the ONUs based on the instruction. An apparatus comprises a receiver configured to receive an instruction instructing a transmit order of transmissions based on optical powers associated with ONUs; a processor coupled to the receiver and configured to process the instruction; and a transmitter coupled to the processor and configured to transmit a transmission based on the instruction.

Optical network power conservation

A method and apparatus for establishing a voiceprint model, a computer device, and a storage medium are described herein. The method includes: collecting speech acoustic features in a speech signal to form a plurality of cluster structures; calculating an average value and a standard deviation of the plurality of cluster structures and then performing coordinate transformation and activation function calculation to obtain a feature vector; and obtaining a voiceprint model based on the feature vector.

Method and apparatus for establishing voiceprint model, computer device, and storage medium

Disclosed are a speech recognition method, apparatus, computer device and storage medium. The method includes: performing a framing and an acoustic feature extraction of a speech-information-to-be-tested according to a default rule to obtain a frame-level speech feature sequence; dividing the frame-level speech feature sequence into n blocks sequentially; inputting all blocks into a preset bidirectional LSTM-RNN model parallelly to obtain an output result of the corresponding neuron in an output layer of the preset bidirectional LSTM-RNN model corresponding to the forward recognition result and backward recognition result of each block to obtain a speech recognition result of the speech-information-to-be-tested. The present application can improve the speech recognition effect significantly and reduce the time delay of the speech decoding effectively.

Speech recognition method, apparatus, and computer readable storage medium

The disclosure relates to technology for signal transmission in an optical communication system. An optical transmitter comprises a directly modulated laser (DML) configured to generate a modulated optical signal in response to a modulation signal. The modulated optical signal comprises a first frequency corresponding to a logical one value in the modulation signal and a second frequency corresponding to a logical zero value in the modulation signal. The modulated optical signal has a modulation symbol rate of 'R'. The transmitter comprises a controller configured to control the DML to establish a target frequency gap between the first frequency and the second frequency. The transmitter also comprises an optical band pass filter (OBPF) coupled to the DML to receive the modulated optical signal and output a filtered optical signal. The OBPF has a 3-dB bandwidth of less than R.

Optical communication system

A speech recognition method, comprising: acquiring speech data to be recognized; extracting a Filter Bank feature and a Mel-Frequency Cepstral Coefficient (MFCC) feature in the speech data; using the MFCC feature as input data of a trained Gaussian Mixture Model-Hidden Markov Model (GMM-HMM) model, acquiring a first likelihood probability matrix outputted by the trained GMM-HMM model; using the Filter Bank feature as an input feature of a trained long short-term memory (LSTM) model which has a connection unit, acquiring a posterior probability matrix outputted by the LSTM model; using the posterior probability matrix and the first likelihood probability matrix as input data of a trained HMM model, acquiring an second likelihood probability matrix; and acquiring a target word sequence corresponding to the speech data to be recognized from a phoneme decoding network according to the second likelihood probability matrix.

Speech recognition with trained GMM-HMM and LSTM models

An acoustic model training method, a speech recognition method, an apparatus, a device and a medium. The acoustic model training method comprises: performing feature extraction on a training speech signal to obtain an audio feature sequence; training the audio feature sequence by a phoneme mixed Gaussian Model-Hidden Markov Model to obtain a phoneme feature sequence; and training the phoneme feature sequence by a Deep Neural Net-Hidden Markov Model-sequence training model to obtain a target acoustic model. The acoustic model training method can effectively save time required for an acoustic model training, improve the training efficiency, and ensure the recognition efficiency.

Acoustic model training method, speech recognition method, apparatus, device and medium

A material for blocking crosstalk, an optical assembly, and a method for preparing the material are provided. The optical assembly includes an optical receive assembly, where a periphery of the optical receive assembly includes a transparent region and a non-transparent region; the transparent region is made of the material, where a first layer of film is located on a side opposite to an optical receiving direction, and a second layer of film is located on a side opposite to the optical receive assembly; and the non-transparent region is of an electrical-signal shielding structure.

Material for blocking crosstalk, optical assembly, and method for preparing material

A framing method and apparatus in a passive optical network (PON) and a system, where the method includes generating a first transmission convergence (TC) frame and a second TC frame separately, wherein a sum of frame lengths of the first and the second TC frame is 125 microseconds (&#x3bc;s), performing bit mapping on the second TC frame to generate a third TC frame, where the bit mapping refers to identifying each bit of the second TC frame using N bits, and sending the first and the second TC frame to an optical network unit (ONU). A line rate corresponding to the second TC frame is lower than 2.488 giga bits per second (Gbps) such that a rate of a receiver on a receiving side is decreased and a bandwidth of the receiver is narrowed, thereby decreasing an optical link loss and increasing an optical power budget.

Framing method and apparatus in passive optical network and system

The disclosure provides a processing method, device and system for data of a distributed storage system. The method includes: acquiring a first schema structure of target data and a second schema structure of source data; and processing the source data into structural data of the first schema structure according to the first schema structure and the second schema structure. The disclosure solves a problem that a data schema structure in a distributed storage system in related arts is inflexible to be described. There is a beneficial effect of improving description flexibility of the data schema structure so as to improve data processing efficiency.

Processing method, device and system for data of distributed storage system

A laser apparatus is provided, comprising a semiconductor substrate, an active layer disposed on the semiconductor substrate, a folded waveguide disposed on the active layer and forming a resonant structure, the folded waveguide comprising at least two substantially straight waveguide portions coupled by a connecting waveguide structure, with the folded waveguide having a first end and a second end located at one or more edges of the semiconductor substrate, wherein at least one of the ends includes a mirror, and an electrode coupled to the folded waveguide and configured to create photons in the folded waveguide when receiving electrical power. The waveguide emits laser light comprising the photons, with the laser light emitted at an edge of the semiconductor substrate.

Folded waveguide structure semiconductor laser

An apparatus comprising a signal generator configured to produce a modulation signal, a filter coupled to the signal generator and configured to filter the modulation signal to produce a cancellation signal, and a reflective semiconductor optical amplifier (RSOA) coupled to the signal generator and the filter, wherein the RSOA is configured to generate an optical signal according to a difference between the modulation signal and the cancellation signal and transmit the optical signal towards a partial reflection mirror (PRM).

Re-modulation crosstalk and intensity noise cancellation in wavelength-division multiplexing (WDM) passive optical networks (PONs)

An optical network system comprising an optical line terminal (OLT) and an optical network unit (ONU) coupled to the OLT and configured to communicate with the OLT via an optical signal. At least one of the OLT or the ONU comprises a closed-loop gain controlled transimpedance amplifier (TIA) comprising a first amplifier configured to receive an input signal, generate a main output signal by amplifying the input signal according to a gain factor of the first amplifier, and generate an auxiliary output proportional to the input signal, an average detector coupled to the first amplifier and configured to receive the auxiliary output, and determine an average of the input signal according to the auxiliary output, and a feedback loop coupled to the first amplifier and the average detector and configured to control the gain factor of the first amplifier according to the average of the input signal.

Closed-loop automatic gain control in linear burst-mode transimpedance amplifier

An apparatus comprising an arrayed waveguide grating (AWG) comprising a plurality of AWG ports, a power splitter comprising a plurality of splitter ports, and a plurality of optical interleavers, each coupled to a respective AWG port and a respective splitter port, for directing incoming optical signals to one of the AWG and the power splitter.

Growing community of inventors

Santa Clara, CA, United States of America

Ning Cheng