It occurs to me that we spend a lot of resources altering our environment and our tools to meet the day to day needs of people with disabilities. This is of course entirely reasonable. Some of these alterations such as wheelchair ramps have become part of our design priorities when we're building new offices or shopping centres. The need is provided for in building regulations and as a society we are conscious of this need. I think in computer terms we have a different opportunity when it comes to access needs; a missed opportunity that can be utilised to the benefit of the perfectly able and those with specific needs. However it requires that interface designers step aside from the usual human perception of ergonomics and approach software design from an imagined perspective of someone who has impaired senses or communication capabilities.
Technologies such as voice recognition and synthesis could form part of an interfaces standard behaviour, it's a faster method of feedback and control for many common operational tasks. We don't treat it as such though, instead it's an option for the aid of people in difficulty and not integrated into our software modus operandi. Interfaces are very static, magnification or other methods of making specific information easier to isolate or consume are hidden away rather than turned into prominent features.
Only in a few situations have technologies used primarily to aid a small proportion of the population been used to build the interface used by everyone. The iPod shuffle's voice control springs to mind. It's part of a very exclusive club however.
I guess I feel that we're missing a trick, we attribute these features to a group of people who require them and somehow fail to capitalise on how these developments can contribute to the rest of us. They don't need to be special 'ease of access' features. Why do we not intuitively assume that when we pick up a new phone we can just ask it to dial the number we want as you would have asked a human operator years ago. These features should be taken for granted as part of the way we interact with computers; for the benefit of the many.