Project Innovations

http url imageInnovation is the raison d’être of the Liberated Learning Consortium. During this project, the team will explore a variety of technical and non technical innovations:

  • advanced multimedia formats for unique user needs
  • novel user interfaces for different disability usage patterns
  • integration of SR with other teaching / learning systems
  • social and peer learning applications
  • new dissemination and communication techniques
  • new training and support methods and resources
  • new partnership models
  • innovative business models to ensure Canadians with disabilities and supporting organizations can efficiently access affordable, reliable, and robust SR systems for improved information access

Recognition Accuracy

Speech Recognition systems rely on statistical data models to perform recognition tasks such as transcription/captioning. The accuracy of a transcript produced by the SR engine will increase greatly if the appropriate acoustic and language models for a given recognition task/domain are present.  HTS currently contains models created for generic transcription tasks. Thus, the models have not been tuned to the university/college lecture environment. 

A higher level challenge is developing transcription capability in new languages. The first generation Hosted Transcription Service contains only a few available languages.  Unfortunately, the system does not currently support Canadian French. 

This project initiates the process of improving recognition performance for the educational domain.  The key “ingredient” to building new models / languages is domain data.  Since this project transcribes lectures for students with disabilities, it will create a rich data pipeline that can be used for modeling purposes.  As the team accumulates more data, they can subsequently develop better tools, which will allow us to eventually capture lectures from other languages.

Collaborative Ecosystem:

Those working in the area of disability understand the reality of working on complex challenges with limited resources. Therefore, leveraging knowledge through collaboration is absolutely essential to achieve incremental accessibility gains.  A single organization cannot possess or develop knowledge in a vacuum.  This project enables a diverse disability community, academic institutions, and industry organizations to collaborate on a shared challenge, and through a model based on consensus and equality. 

Because the project is cross-disability based, it promotes dialogue between diverse disability groups and synergies with other accessibility efforts.  The structure and philosophy of this project supports an overarching objective of engaging persons with disabilities as research partners rather than subjects. A key goal is to increase the aperture of this research by engaging new organizational partners.