Related Experiment Videos
User performance with speech recognition: a literature review
1Center for Ergonomics, University of Michigan, Ann Arbor, Michigan 48109-2117, USA.
Assistive Technology : the Official Journal of RESNA
|January 18, 2003
Summary
Speech recognition technology offers computer access for people with disabilities. However, user performance, especially for those with physical disabilities, significantly lags behind system potential, highlighting a need for further research.
Area of Science:
- Assistive Technology
- Human-Computer Interaction
- Disability Studies
Background:
- Speech recognition systems are increasingly used for computer access by individuals with disabilities.
- Understanding user performance is crucial for effective implementation and accommodation.
- Existing research provides a foundation, but gaps remain, particularly for users with physical disabilities.
Purpose of the Study:
- To review current knowledge on user performance with speech recognition systems.
- To focus on the application of these systems for accommodating physical disabilities.
- To identify barriers and propose future research directions.
Main Methods:
- Literature review of studies on speech recognition user performance.
- Analysis of performance data, focusing on text entry speeds.
- Examination of factors contributing to the performance gap.
Main Results:
- Current speech recognition systems have a theoretical text entry speed of 150 words per minute.
- Actual user performance in real-world applications ranges from 8 to 30 words per minute.
- Significant barriers exist, including the time and effort required to correct recognition errors.
Conclusions:
- A substantial performance gap exists between the potential of speech recognition and actual user achievement.
- Barriers such as error correction costs limit the effectiveness of these systems.
- Further research is critically needed, especially studies involving users with physical disabilities to improve assistive technology.