There's no There There When Talking to Machines
Talking to Machines (TTM)
Truly interactive speech is far more complex than automatic speech recognition (ASR) and requires additional solutions. We have 30+ years of experience in speech technology to share.
Automatic speech recognition (ASR) is not in itself enough to develop productive interactions for talking to machines.
TTM today is mostly used for only a few set of tasks like setting a timer, controlling music, starting an Internet search to name a few.
Human speech interaction is far more complex than today's smart speakers and voice assistants.
Humans do not mistake a car horn for speech.
Learn how to address this error.
Automatic Speech Recognition (ASR), turning sound into text, is based on probability and statistics. Humans rarely talk for more than four seconds without taking a breath. We show how this is important in speech recognition. More than ASR is required to be certain it heard ‘Yes’ or ‘No’. We delve into what is needed to correct for errors.
