The problem with current speech recognition software (and why it is still maddeningly inaccurate sometimes) is that it cannot learn and receive immediate feedback from users as occurs in natural speech. Now, computational linguists are developing new approaches which allow this to happen, leading to much higher accuracy rates.
- Talking To a Machine: Top Speech Recognition Tools (homework.answers.com)
- Microsoft’s new xbox speech recognition like siri (technologynewsworldwide.com)
- Microsoft’s new Xbox will include improved Siri-like speech recognition (theverge.com)
- Speech Recognition Software in the Legal Field (lawprofessors.typepad.com)
- Google Research: More (big) data means better speech recognition, by @DerrickHarris (nextlevelofnews.com)
- Accessible Learning – UDL (slideshare.net)