The popularity of Open educational resources (OERs), Massive Online Open Courses (MOOCs) and OpenCourseWare (OCW) is growing exponentially. Hundreds of millions of learners are accessing educational multimedia through platforms such as Apple’s iTunes U, YouTube EDU, EdX and Coursera. While the use of multimedia in these offerings provides numerous educational benefits, it can also introduce accessibility barriers. Ensuring that multimedia content is published with captions and/or interactive transcripts is a key challenge facing educators.
The overarching objective of this initiative is to introduce the use of speech recognition to OER / OCW / MOOCs to produce accessible educational content. In addition to increasing awareness of the need for accessible multimedia among key stakeholders, the Consortium will actively utilize Speech Recognition in real world OpenCourseWare / MOOC environments.
The University of Massachusetts Boston is delivering a Massive Open Online Course (MOOC) titled Molecular Dynamics for Computational Discoveries in Science.
This MOOC introduces two unique approaches to OpenCourseWare. First, the entire course will be captioned using speech recognition technology. Course media will not only be fully accessible, but the case study will additionally allow researchers to gather real-world data about the application of Speech Recognition to MOOC development:
- how is speech recognition integrated into a captioning workflow for a typical MOOC?
- how effectively do existing Hosted Transcription Systems perform captioning tasks?
- how do costs of using speech recognition compare to traditional captioning costs?
- how do MOOC learners utilize captions / interactive transcripts?
UMASS Boston’s Center for Innovation and Excellence in eLearning will record a series of course modules, which will be transcribed using Nuance’s Dragon Server SDK transcription feature. The SDK features an “eyes free” enrollment tool that allows researchers to create a speaker dependent profile for the instructor using actual transcribed course material. As course media is produced and transcribed, the synchronized transcript/media will be used iteratively as training data. This approach has been successfully used to dramatically reduce Word Error Rates during transcription and reduce demands placed on the instructor.
The course will also showcase Synaptic Global Learning’s AMOL (Adaptive Mobile Online Learning), which will make this the first truly adaptive MOOC in the short history of MOOCs. The platform allows for the coursework to adapt to each individual’s learning style.
Partners from the UMass Boston Ross Center for Disability Services and Saint Mary’s University will provide research and technical support.