[ad_1]
SINGAPORE – Singapore's mobile operators receive almost 200,000 calls a year, every minute is vital.
In order to lighten their workload, the Singapore Civil Defense Force (SCDF) and four other government agencies are turning to artificial intelligence (AI) and have developed a voice recognition system able to transcribe and record each call received in real time – even if it is in Singlish
For now, the system is programmed to recognize English and Mandarin with some Hokkien. and Malay, although it can be customized to incorporate others.
AI Singapore (ISGF), a program of the National Research Foundation, invests $ 1.25 million to set up the AI research laboratory that developed the system
claims to have created the first code-switch or mixed-language speech engine
It was developed using artificial intelligence, such as in-depth learning technology that works from algorithms. "This will improve the distribution of SCDF's emergency medical resources and improve the overall health outcomes of those in need," said SCDF's Director of Operations, Assistant Commissioner. Daniel Seet
He added that this would help reduce the time that dispatchers in the 995 SCDF operations center have to record in the information.
"In an emergency, every minute counts," he added
The AI Lab is led by Professor Li Haizhou, an expert in speech therapy, text and natural language. from the National University of Singapore and Associate Professor Chng Eng Siong from Nanyang Technological University
They have been working on speech recognition for a decade or so. [19659002] Prof Li said that such a code switching system is currently not commercially available.
He said, "This technology works better than commercial engines because it can accurately recognize conversations with words of different languages." To solve this problem, Singapore has collected more than 1,000 hours of samples from Singapore and Penang – a state that mixes languages similar to those of Singapore – as well as Singaporean recordings of radio stations, YouTube and SoundCloud
These recordings are transcribed manually in the text, then the system "learns" the Association between text and speech samples collected
The system has "learned" about 40,000 English and Mandarin words each, and has an accuracy rate of about 90 percent. [19659002] Unique words that the system can recognize include "jiak ba bueh" and "hoh boh" – "did you eat" and "how are you" in Hokkien – and local dishes such as char kway teow and nasi lemak
The lab is comprised of five IA engineers and located at Innovation Building 4.0 of the Kent Ridge campus of the National University of Singapore and became operational from July 1st.
Professor Ho Teck Hua, executi He added that the system could also benefit businesses because it can be customized according to their business needs.
million. Tan Kok Yam, deputy secretary of the Smart Nation Agency and Digital Government, said: "The government is eager to harness artificial intelligence to better serve our citizens .GovTech is collaborating with AISG to develop solutions that can improve planning and service delivery. "
Dr. Manjunath Bhat, Research Director at Gartner Consulting and Research, said," Multilingual transcription will make life easier for older people. " Even though the communication systems are going from analog to digital, the human language itself remains analog. The new solution allows computers to speak in the language of the ordinary person as opposed to learning the language. l & # 39; adaptation. to digital interfaces. "
The details of when, where and how the system will be tested have not been announced yet.
[ad_2]
Source link
