Hackernoon logoAutomated Speech Recognition Increases Productivity In Singapore's Government Services by@markli

Automated Speech Recognition Increases Productivity In Singapore's Government Services

Author profile picture

@markli

Mark Li is a persona created to compile the brains and thoughts of professionals at Sentient.io

The Singapore government launched its Smart Nation Initiative in November 2014. Since then, Singapore has supported the extensive use of technology to enrich the lives of its citizens. Artificial Intelligence (AI) solutions have been successfully harnessed to improve the efficiency and quality of services. 

Automated Speech Recognition (ASR), which is the technical name for text-to-speech conversion, has become one of the most widely used applications of AI, so much that a special flavor of ASR was created just for Singaporeans.

In this article, we will explore the multiple ways in which Automated Speech Recognition (ASR) has provided value to different organizations, specifically Singapore government services. We will also look at how Sentient.io’s Voice AI and text-to-speech services were instrumental in the implementation of Automated Speech Recognition (ASR) technology for government services.

The Hidden Worth of Automated Speech Recognition

While most people recognize Automated Speech Recognition as the technology that makes funny conversations with Siri, Google, and others possible, it is so much more than that.

In the services industry, Voice AI-based bots are a boon because they always manage to sound pleasant and patient. Moreover, they not only understand spoken words but also the context in which they are spoken. 

In addition to interacting with customers/users, ASR can also be applied in the workplace. ASR-enabled applications can listen to conversations or speech and generate transcripts that can be used for in-depth analysis. Transcribed or text-based records are not just easy to maintain, they also provide easy access to information via digital search. 

Some distinct advantages of using automated speech-to-text services are as follows.

  1. Transcribing speech to text manually is a time-consuming and tedious activity requiring hours of human effort. Automation ensures that this time can be put to better use.
  2. Automation reduces the chances of errors or omissions during the transcription process
  3. When it comes to sensitive information, it may not be possible to outsource the transcription task to external contractors due to security or privacy concerns.

A well-tested ASR service that is also sensitive to local nuances in speech can help to generate transcripts with high accuracy. There are many processes in both the public and the private sector that can greatly benefit from the power of ASR.

The following are some examples of where Speech-To-Text services can help in better governance.

  1. Governing bodies need to conduct meetings daily. Accurate and easy to access verbatim records of such meetings would ensure that significant information is always available. This is especially beneficial for high-level meetings or discussions that can affect laws, regulations, and policies.
  2. When it comes to court rulings, lawyers often rely on the records of past judgments to build their case. Such records can be easily generated using ASR.
  3. Parliament proceedings, speeches, and debates that are fully recorded can be transcribed for public consumption. In situations where recordings are available on video, the same service could be used to include subtitles in the video.
  4. “This conversation may be recorded for quality assurance” - That's right! Call centers usually record conversations. Analysis of transcripts of such conversations can help to address citizen pain points when interacting with government bodies.
  5. Interviews for new employee positions may be recorded and transcribed for reference in the future especially for qualified candidates who were not recruited.

To enable the easy use of ASR in any locale it should be tailored to the specific nuances of the language used in that locale. Let us now see how this factor is relevant to ASR technology used in Singapore.

Why and How did Singapore create its own flavor of ASR?

English spoken in Singapore can be standard Singapore-English or the colloquial form, also known as Singlish. Any speech-to-text engine built for Singapore should consider nuances that are unique to Singaporean English. Following are some of the words and terms we use in our day-to-day life which are not pure English and would not be recognized by any other ASR.

  1. Words like “ang-moh” and “kiasu” are commonly seen in news articles and tv shows.
  2. Phrases like “double confirm” and “act blur” may sound English but are only unique to Singapore.
  3. Terms like “4D” (the lottery system) and “O$P$” would only be relevant to Singaporeans.
  4. Slang words like “Chope” which is also the name of a local website have become a part of a normal Singaporean conversation.

Existing speech technologies were not capable of transcribing locally accented English accurately. To achieve high accuracy, Sentient.io trained its Automatic Speech Recognition model on IMDA’s National Speech Corpus (NSC). This corpus contains 1 Terabyte of annotated speech audio. This ensures accuracy in transcription when spoken English is accented or colloquial. Let us now see how this technology helped some of the government services to build ASR based solutions.

Singapore Government Services That Benefitted From ASR

Sentient.io’s expertise in ASR technology has enabled several Singapore government agencies to build robust solutions using ASR. Following are some agencies that have benefitted from Sentient.io’s ASR solutions.

  1. Singapore’s largest content creator and national media network for both radio and television which produce all kinds of content for TV and radio, including dramas, news, talk shows, reality tv, documentaries, etc. Sentient.io’s ASR has helped them to improve their productivity in News Reporting by 500% and Drama Subtitling by 600%.
  2. The Ministry that oversees information and communications technology, the media and design sectors, public libraries, as well as the Government's information and public communication policies engage Sentient.io’s ASR to help with the transcription of the Committee of Supply 2020. This included a transcription of speeches by ministers and senior parliamentary secretaries who participated in the event.

How can you benefit from Sentient.io ASR?

Sentient.io not only possesses the expertise in developing ASR solutions for government services, but the Sentient.io in-house team can also quickly customize their ASR services for a broad range of industries and use-cases.

Also published at https://medium.com/sentient-io/automated-speech-recognition-helps-singapore-government-services-to-increase-productivity-83b39d59d0e3

Read my stories

Mark Li is a persona created to compile the brains and thoughts of professionals at Sentient.io

Tags

Join Hacker Noon

Create your free account to unlock your custom reading experience.