Semantic Transcription Technology for iOS, OSX and Linux

1. Introduction to Semantic Transcription.

With the rising prominence of vocal interaction systems in both work and lifestyle, the derivation and representation of intent-meaning (of verbalized content) has become a core component of such systems. Through semantic transcription, vocal engagement systems can comprehend user intent and so react and respond in an appropriate manner, assuming the systems knowledge domain covers the semantic content.

Semantics and Intent are related but not equivalent. They are related in the sense, that the Semantic is generated from the sentence grammar and Intent is derived from the Semantic. But they each encode into very different logical sub-units. The semantic encodes to the framing, expression and content (content being the subjects and actions). Intent encodes to a symbolic-programmatic form that represents the intended interactions between subjects and actions. Also, the semantic has a more linguistic focus, encoding both expression (how it was expressed) and meaning whereas intent is agnostic to how the sentence was expressed and is only concerned with subject-action interaction.

Originware's Semantic Transcription system is an SDK for use in systems engaged in or dealing with vocal interaction. The SDK engine accepts input text (a single sentence) and transcribes to the semantic. There is additional response resolution support, in that the semantic can be routed and passed through the various responders which delineate the individual knowledge and capability domains (that the application supports), to determine a "matching" responder. That responder through application support maps the semantic to a custom application intent which is then evaluated to determine the appropriate vocal-textual responses and actions.

The SDK is applicable to the full vertical stack, from SOC (embedded IOT devices) to mobile app, to the hosted server, to node swarms. Transcription is extremely performant (see the Performance Benchmarks video) with a compact memory foot print (on Linux of less than 45M). It suitable for 32bit CPUs and beyond.

2. Semantic Transcription Videos.

3. More information.

Please read the product preview: The Semantic Transcription Technology Product Preview

You can also contact Terry Stillone  (terry@originware.com) for more direct information.

4. Product Developmental State.

The SDK is currently in early beta, migrating from version 0.9 to the target product version of 1.0 with the production target version as 1.1.

Home

About

Contact