Learn how Digital Service Lab’s tools and technologies have been propagated among the services sector to catalyse digitalisation in Singapore.
National Speech Corpus
Chatbot demo integrating ASR and TTS software on show at IMDA SG:D Industry Day
The chatbot, SG Restaurant, is able to accurately take spoken orders and read back orders in a Singaporean accent.
NSC was used by Sentient.io's AI and Data Platform to make building applications easier for developers
The NSC dataset has been used by Sentient.io to train the Automatic Speech Recognition (ASR) microservices, enabling it to recognise and transcribe Singapore accented speech.
NSC enabled Sentient.io's AI speech microservices to understand Singapore accents and speech more accurately than Siri and Alexa
Sentient.io's founder shares about how their natural language processing (NLP) draws on NSC to understand and transcribe Singapore speech more accurately than Siri, Google Home and Alexa.
DSL's project collaborations with Sentient.io was featured in ST
DSL collaborated with Sentient.io to develop their defining products: chatbots recognising local lingo for banks, and subtitling services for media companies.
NSC enhanced with 1,000 hours of natural voice recording
Straits Times' report on NSC V2.0 which added 1,000 hours of Conversational Speech, an update of NSC V1.0 which comprised of 2,000 Read Speech.
The National Speech Corpus was presented at INTERSPEECH 2019
IMDA DSL’s “Building the Singapore English National Speech Corpus” paper was accepted and presented at the INTERSPEECH 2019 conference held in Austria.
OCBC’s collaboration with IMDA
Head of E-Business of OCBC Bank, Mr. Pranav Seth shared how OCBC was working with IMDA to provide a uniquely Singaporean voice for their chat-bot, “EMMA”.
Click here to watch the video
Artificial intelligence library with voice sample in Singapore English to launch in 2018
Straits Times’ coverage on the launch of IMDA’s AI library with voice samples in Singapore.
Singapore embraces AI with open source libraries and talent development
IMDA’s announcement on initiatives in Artificial Intelligence (AI), consisting of the National Speech Corpus V1.0, Intelligent Sensing Toolbox and the AI Talent Development programme.
NSC is used by AISG in the development of SCDF Automatic Speech Recognition (ASR995) engine
The NSC has been used by AISG to develop the Singapore Civil Defence Force (SCDF) Automatic Speech Recognition engine (ASR995) to understand locally accented English.
Natural Speech and Transcription Technologies
New initiatives launched to help companies adopt emerging technologies
Channel News Asia’s coverage on the release of IMDA’s National Speech Corpus. The NSC is used to teach an automated transcription software co-developed by Mediacorp and IMDA.
Click here to watch the video
Intelligent Sensing Toolbox
Chiller Doctor V2.0’s soft launch at IMDA SG:D Industry Day
The Chiller Doctor analyses data from sensors and operation crew reports to detect faults early, reducing risk of chiller failure.
IMDA announces collaboration with BCA’s SLEB Smart Hub to develop AI-driven chiller fault detector, Chiller Doctor
The Chiller Doctor currently in it’s early pilot stage is expected to perform predictive detection of faults, minimise reliance on system maintenance and optimise energy conservation in buildings.
Intelligent Data Centres
The Digital Services Lab (DSL) Team designed outlier detection to help National Supercomputing Centre (NSCC) identify the causes behind abnormal behaviours of server nodes within a data centre.
For this project, large amounts of data sets from thousands of computer nodes were analysed. NSCC’s live power usage effectiveness (PUE) readings improved from 1.32-1.36 to 1.28-1.29.
This project was piloted successfully in 2017, in collaboration with Nanyang Technological University (NTU) and Red Dot Analytics (RDA).
The Digital Services Lab (DSL) Team replicated the Intelligent Sensing model to help Venture Corporation improve their printed circuit board assembly (PCBA) manufacturing process.
For this project, millions of machine log entries were captured and analysed to predict problematic component feeders so as to reduce the chances of failure.
The Digital Services Lab (DSL) Team has worked with Keppel Data Centres to highlight the benefits of Machine Learning-based Outlier Detection in diagnosing failures and detecting false alarms.
For this project, the health conditions of Chiller Plant Systems including Chiller, Water Pumps, Cooling Towers, Air Handling Units (AHU), were analysed in order to identify anomalies. The baseline was established by using machine learning algorithms modeling normal and various fault conditions to detect deviations, that could signify equipment wear-and-tear.
A smart future for data centres
Artificial intelligence and how it can help to boost the efficiency of data centres were discussed in this dialogue session.