National Yunlin University of Science and Technology

National Yunlin University of Science

Jeng-Shin Sheu

Jenn-Kaie Lain

Sin-Hong Lyu

Cheng-Hsuan Lu

The invention is a speaker identification system, which is provided to train a speaker model based on a meta-learning approach. Through the training of a plurality of episodes, the speaker model is updated by backpropagating the gradients of a composite objective function comprised of two loss functions and each episode consists of a support set of long utterances and a query set of short utterances. With this single speaker model, the invention converts an input utterance into a speaker embedding vector. This enables the speaker identification system to identify different enrolled speakers solely through the comparison of speaker embedding vectors, effectively blocking spoofing attacks and impostor intrusion. Consequently, the invention is characterized by its lightweight nature, real-time response, suitability for short utterances and open set environments, and can be implemented using low-cost embedded hardware.

Real-time speaker identification system utilizing meta learning to process short utterances in an open-set environment

A network multiple-input multiple-output (MIMO) wireless signal transmission and power control system includes a plurality of adjacently arranged cells. Each cell is a regular polygon and includes a plurality of adjoining sides, a base station located at a center of the cell, a plurality of directional antennas equipped at the base station, and a power adjustment unit electrically connected to the directional antennas. The directional antennas point towards connection points of the adjoining sides, and sectorize the cell into a plurality of kite-shaped sections by regarding the directional antennas as centers. The power adjustment unit controls power output of the directional antennas. By changing pointing directions of the directional antennas, the cell can be sectorized into the kite-shaped sections to significantly increase system capacity and communication quality.

Network multiple-input multiple-output wireless signal transmission and power control system

A channel estimation method for an OFDM relaying system, which is based on an EM algorithm and applied to an AF OFDM relaying system, wherein channels from a source node to a relaying system and from the relaying system to the destination node in a wireless communication system are respectively estimated at the destination node. The channel estimation method of the present invention comprises steps of: constructing a system model; setting an expectation function; performing maximization processing; and performing iteration. The method of the present invention respectively estimates the channels from the source node to the relaying system and from the relaying system to the destination node, whereby are optimally combined the signals of the direct path (source-to-destination) and the signals of the relay path (source-to-relay and relay-to-destination) at the destination node.

Channel estimation method for OFDM relaying system

An OFDM-based relay-assisted channel estimation method enables a composite channel (source-relay-destination) to be disintegrated into a source-to-relay channel and a relay-to-destination channel and separately estimated at the destination node in a relay-assisted way. A relay-amble signal is added to a sub-frame when the relay node is forwarding the amplified signals from the source node. Through the relay-amble signal, the second channel estimation from the relay node to the destination node is performed to obtain a second channel coefficient, and then uses an ML (Maximum Likelihood Estimator) algorithm to estimate a first channel coefficient from the source node to the relay node. After having estimated the coefficients of the two channels, the optimum diversity combining is performed at the destination node to improve accuracy of received data.

Growing community of inventors

Yunlin County, Taiwan

Jeng-Shin Sheu