Logo

Taiwan scholars reduce distracting noise from hearing aids using deep learning

Taiwan scholars reduce distracting noise from hearing aids using deep learning

2022-12-02

A team of Taiwanese scholars has developed a big upgrade to hearing aid technology. It’s the first system to cut confusing noises from hearing aids using an AI deep learning. The computer scientists from Academia Sinica hope their innovative tools will make hearing aids much more effective, and lead to easier communication for people with hearing problems.

When a person’s hearing is damaged, they may rely on artificial hearing aids for daily communication. But environmental noises can make it hard to distinguish sounds. Now this team from the Research Center for Information Technology at Academia Sinica has developed an “AI Oral Communication Support Technology.” They hope to incorporate it in hearing aids and electronic earpieces, and make sounds clearer.

Tsao Yu
Academia Sinica researcher
This issue has been around a long time. Only in 2013 did we start to have lots of data, big data. Then we had big models, and our hardware modeling got faster. Then this tool became really popular, and it got really effective. We are the first team in the world to use deep learning to enhance sounds.

The team says that their algorithm for enhancing speech, based on deep learning methods, can reduce distracting sounds like car noises and music when someone is trying to have a conversation. It also uses an AI speech enhancement system. You can use your phone to clarify sounds that are unclear, making communication easier.

Lin Kuang-hung
National Science and Tech. Council
The research results have received a high level of recognition and support, in Taiwan, abroad, from academia and from industries. We want to bring this into practical application.

The research has won the 2022 Future Tech Award. It’s also been published in more than 50 papers in international journals. The team wants to collaborate with businesses worldwide to bring their work to those who could use it.

開發AI口語溝通輔助技術提升國內智慧醫療

2022-12-02

為提升國內智慧醫療的競爭力,國科會補助中研院,開發"AI口語溝通輔助技術",利用AI技術協助溝通或幫助聽障者提高學習品質,目前技術已通過臨床試驗,並獲包括2022未來科技獎,也希望未來能跟國內外業者合作,導入產品開發。

有些人因聽覺受損,必須靠人工耳和助聽器來解決日常溝通障礙,但常見的環境噪音經常造成語音品質降低。為了提升語音品質,中研院團隊研發出, "AI口說溝通輔助技術",將AI語音技術導入包括助聽器、人工電子耳等聽覺輔助工具,要讓聽損者聽得到,也聽得懂。

[[中研院資訊科技創新研究中心研究員 曹昱]]
“這個題目其實早又有了,只是在2013年開始,我們有大量的data,我們有big data,然後我們有很大的運算,我們的硬體運算加速了,這個工具就變得非常熱門,然後效果變得非常好。我們是全世界第一個用深度學習,用到語音增強。”

中研院團隊提出"基於深度學習的語言增強處理演算法",可應用在口語溝通障礙者對話時,能消除車聲、音樂聲等環境噪音,讓他們清楚聽到對方的聲音。另外,也將"利用AI的基礎語音增強系統",透過手機,將患者不清楚的語音轉換成正常語音發出,提升溝通效率。

[[國科會主任秘書 林廣宏]]
“研究成果獲得國內也好、國外也好、學界跟產業界,高度的肯定跟支持。希望可以(盡快)落實在我們生活運用上面。”

目前該研究成果已獲2022未來科技獎,相關研究已發表超過50篇國際期刊論文,未來更希望與國外業者合作,幫助更多有溝通障礙的人。

Related News

KMT nominates Taipei Councilor Wang Hung-wei to run in Taipei legislative by-election

2022-12-02

Speculation mounts over who will be the next DPP chair

2022-12-02

China Medical University shows off development and various inventions at Healthcare+ Ex

2022-12-02

Taipei to lift 44-year-old scooter ban on Zhongxiao West Road starting Dec. 14

2022-12-02

Taiwan holds regional qualifier for International Super Model Contest

2022-12-02