Recently, Liu Jinjiang, a PhD candidate of Prof. Zhang Xueliang-led research team, won 2 silver prizes, on behalf of IMU, respectively in “Hearing Industry Research Consortium Prizes Objective Evaluation” and “Hearing Industry Research Consortium Prizes Listening Test” in the 2nd Clarity Enhancement Challenge.
With the aging of population, the number of people with hearing impairment will increase sharply. According to the statistics and prediction of the Lancet, 1.5 billion people around the world had got hearing impairment in 2019 and the number will increase to 2.5 billion in 2050. As a result, there is a huge demand for hearing aids. The Clarity Enhancement Challenges are jointly organized by such British Commonwealth universities as University of Sheffield, University of Nottingham, University of Salford and University of Cardiff and such industrial partners as the Hearing Industry Research Consortium (IRC) and Amazon for the purpose of clarity enhancement. The challenges mean to organize the open evaluation of the algorithm for the hearing aids and encourage the innovative machine learning methods to address the noises of hearing aids. The 2nd Clarity Enhancement Challenge has attracted participants from such universities and academic institutions as Carnegie Mellon University, Tokyo Metropolitan University, University of Sheffield, University of Oldenburg, National Taiwan University, Academia Sinica and Università Politecnicadelle Marche and such foreign hearing aid companies as 4Pulse Audition and Horizon Robotics. The team from Carnegie Mellon University won the gold prize.
In this clarity enhancement challenge, Liu Jinjiang proposed the strategy of low-latency STFT for low-latency senarios. And the latency of the time-frequency-domain clarity enhancement algorithm system based on the strategy, depends on frame shift and his strategy solves the problem of the application of high-resolution frequency-domain neural network in super low latency scenario. Liu proposed such innovative high-performance neural networks as DRC-NET(CEC2) and ICSE(CEC ICASSP, 2023) for clarity enhancement of double ears through multi mikes to realize the enhancement of beams of directed end-to-end voice of double ears under ultra signal-to-noise ratio.
Liu Jinjiang is a third-year PhD candidate of Prof. Zhang Xueliang’s research team and focuses on the research of multi-channel clarity enhancement. Liu has published 5 articles in ICASSP and INTERSPEECH, top conferences in the field. He has also got involved in the pre-research and solution of the key issues in the industry such as multi-channel noise reduction, elimination of reverberation, echo cancellation in acoustics and clarity enhancement of hearing aid of double ears.