There are existing transcribing/translation software programs that have limitations such as the type of platform they can run on, costly licence, inability to interpret Singlish and dialects, as well as inaccuracy in transcribing long speeches. To address the above limitations, we have built a low-cost multilingualspeech-to-text (S2T) transcription and translation software that runs on common mobile devices, such as smart phones, tablets, laptop computers and smart glasses. The S2T transcriber/translator is a low-cost assistive tool that facilitates more effective communication for the deaf and elderly folk in public. It is created with a localised context that is able to handle different accents in Singapore.
S2T transcriber/translator has the following features: Recognise up to 81 different languages, such as English, Chinese, Malay, Tamil, and also some colloquial Singlish Allow the transcription of speech to be conducted in a silent environment, a location with an average level of noise such as murmuring, or a noisy environment Provide language translation capability. It permits users to select a specific language to be translated, with an audio output option. Provide non-stop operation (up to 3 hours) for long conversations Management and support of multi-language libraries are entirely handled by the server, thus making constant software updates to the end device and extension oflanguage libraries more practical and manageable
The S2T transcriber/translator leverages on the mobile/wearable computing devices to provide a more effective real-time transcribing tool for the deaf and the elderly to interpret spoken words. However, the software can be adapted to meet various market opportunities, for example: Patient care in hospitals Elderly care services Customer care in retail outlets Notes taking for students and secretaries Language translator for travellers Provide captions for radio, live TV broadcasts, movies and digital media files Virtual assistant & voice automation
S2T transcriber/translator is: Low cost Processes data faster and more accurate than other existing technologies Able to tolerate local accents and surrounding noise Able to provide non-stop speech-to-text transcription and translation Supports many-to-one communication in open space or indoor environment Recognises common languages that are spoken in our multi-ethnic and multi-cultural society