Description:

 

Many communities in New Zealand are quite vulnerable to tsunamis and/or earthquakes.  During these disasters few people know what to do. A potential solution is a very local (LPFM) radio station that:

• Always broadcasting so the community becomes familiar with it, it is known to be functional, promotes preparedness and just generally works for the community engagement.
• Battery/solar powered (for when power fails)
• Able to be switched to emergency message mode easily and remotely - with situation messages preloaded.
• Able to be updated remotely through a variety of channels by various authorised people (avoid single points of failure).

The system in this solution can receive text content through various channels and automatically broadcasts this as speech. This has been set up in a small Wellington community. The problem is that the text to speech system was not developed with the New Zealand context, and cannot pronounce New Zealand place names well, particularly Māori names. This creates intelligibility problems which compromises the effectiveness of the emergency broadcast. This project will in involve creating a robust prototype system which will enable emergency broadcast for a New Zealand environment, it needs to be able to operate stand-alone and be cloud/internet based. The project will involve research into embedded systems, text analysis, speech synthesis, and synthetic voice creation. It will utilise existing New Zealand English and Māori pronunciation lexicons we have already developed.

Type:

Undergraduate

Outcome:

Prerequisites

None

Specialisations

• Computer Systems Engineering
• Software Engineering

Categories

• Embedded Systems
• Image & Voice Processing
• Smart Phone & Tablet Applications

Supervisor

• Catherine Watson

Co-supervisor

• Jesin James

Team

• Riley Fromont
• Alexander Monk

Lab

Lab allocations have not been finalised