The Future of Voice Arrives
The VoiceBox Conversational Search Platform™ makes applications and devices understand, not just listen.
For the VoiceBox Conversational Language Processor (CLP) and the VoiceBox Voice Search Engine (VSE), the analogy of hearing vs. understanding applies. If the ASR is the ear, then VoiceBox is the brain. Humans find relationships in keywords and environmental clues. They combine all this knowledge to infer understanding for what they’ve heard, then employ Cooperative Conversations™ to clarify actual comprehension and action.
Dig deeper into VoiceBox Technology Architecture; click on the diagram
below:
Defining
Conversational Voice Search™;the catalyst for emerging markets
Conversational Voice Search is the ability to search and navigate content and services from any IP device or network, by using the natural, free-form language and the casual speech of human beings. Unlike traditional speech solutions, with preset commands or word orders to memorize, VoiceBox Led™ applications work by understanding casual conversational language, even when there is noise, pauses, and human stuttering.
The
Challenge: Humans vary drastically in the use of vernacular and style
of requests.
Variables such as stress and distraction affect what users verbalize. Words out of the expected vocabulary, stammering, pauses, noise, accents and quality of audio paths can cause requests to be distorted or confused. VoiceBox uses context and user intent to correct such “mistakes” and will often generate a correct query under the most difficult of circumstances. As a result, users of integrated devices in the car, on the mobile phone and in home networks can speak freely and naturally.
The
VoiceBox Way: Getting action through Cooperative Conversation™
Cooperative Conversation assumes casual speech and tolerance of noise and other speech ambiguities. Humans find relationships in keywords and environmental clues. They combine all this knowledge to infer understanding for what they’ve heard, then employ Cooperative Conversations™ to clarify actual comprehension and action.
One of the crucial missteps in a vast number of today’s speech interfaces is that they either “dumb down” requests or they require significant learning on the part of the user. Instead of reacting to keywords, VoiceBox examines the relationship of keywords across multiple utterances. This shared knowledge is used to establish contextual clues to aid in the understanding of the user’s request. The VoiceBox Cooperative Conversational approach assumes imperfect input and attempts to handle misunderstandings in the most human-like way possible.
VoiceBox
creates a human-like experience in several ways:
- Utilizes Shared Knowledge of the conversation to dynamically establish context
- Creates Adaptive Responses that enable the machine to make assumptions depending on confidence or interaction with users “in-context”
- Applies adaptive misrecognition mending to sharpen sentences and correct misrecognized words.
The goal is to enable a faster and more efficient Human Machine Interface (HMI), make conversational voice interactions with devices “normal” and give beginning to advanced users the power to access and navigate more information, services and features.
