Smart TV Voice Recognition
2016 was a big year for the Voice Recognition feature in Samsung Smart TV as the feature went through a massive revamp. There were many VOCs (voice of customer) from 2015's Voice Recognition and was time for some major changes. First one was the redesign of the voice indicator, in which I was responsible for the visual interaction design work. To achieve for best synchronization and accuracy, I collaborated with developers to implement the design as code. Second was the change in layout and position of the feature. As TV experience is primarily about watching the content, we made sure that the Voice Recognition feature was not distracting this core experience, while making sure that the feature was clear enough. Last was a lot of the technical updates that were done with the domains and implementing ASR (automatic speech recognition) technology, so that user input is translated from speech to text in real-time.
Problems with 2015 Voice Recognition
Voice Agent not very responsive
Since the voice indicator was developed with series of PNG images, the indicator was not accurately in-sync with user's voice input.
Limited in functionality and capabilities
Features such as changing the volume level and turning on/off certain mode were very limited in terms of its capabilities.
Problems with layout and composition
As the area for voice recognition feature was located at top of the screen, it visually looked out of balance and with no dark background behind the interface, its legibility against content in motion decreased.
Difficult to know voice input status
Since ASR technology was not offered, user's voice input was unable to be translated from speech to text in real-time. This caused difficulties in whether the system accurately recognized the voice input or not.