Jumper

Modules

• Speech-to-Text
• Entity Linking
• RTS-Knowledge base population
• RTS-Knowledge visualization

Speech-to-Text

Currently using browser APIs for STT. Exploring alternatives like Watson. Would like to be platform agnostic.

Quality of service takes precidence over others.

• Web Speech APIs
• Microsft Cognative Services - In JavaScript Browser
• We also have self hosted services like Moz-DeepSpeech.

We can fine tune the grammer of the Speech Regonition system to focus on Entites? JSgf - JSpeech Grammar Format

Speech recognition systems provide computers with the ability to listen to user speech and determine what is said. Current technology does not yet support unconstrained speech recognition: the ability to listen to any speech in any context and transcribe it accurately. To achieve reasonable recognition accuracy and response time, current speech recognizers constrain what they listen for by using grammars.

Entity Linking

• http://nlpprogress.com/english/entity_linking.html
• opentapioca - Live Demo
• NIF = NLP Interchage Format - NIF 2.0 Specs

Opentapioca can be self hosted, and also has a public NIF API

RTS-Knowledge base

• Time
• What / Where / When / How
• Its like a memory

Visualizations

• Do we define a information min size?
• Do we define a information max size? (eg. at max only at a go n entities are added to the graph)