Voice-driven applications are now ubiquitous, as is evident from the rise in popularity of virtual personal assistants (Siri, Cortana, Alexa..) and voice activated in-car command systems. These applications use Audio Speech Recognition (ASR) technology to decipher speech from the audio waveform. ASR performs very well in ‘clean’ environments, but word accuracy generally degrades in more real-world, noisy environments. LipRead VSR is audio noise agnostic and can combine with the ASR to improve overall system performance in situations where background noise exists and where a camera can be trained on the head of the speaker.
Liopa has developed a series of patent-pending techniques to dynamically integrate 3rd party ASR solutions with LipRead VSR giving optimal accuracy across varying levels of audio & video noise.
Liopa’s AVSR solution can truly unlock the potential of the Voice-first interface.