Take an example of a "Speech to Text" Software installed on a PC, these have several limitations and a fuzzy learning process. It works with people with known dialects, good pronunciation and proper language syntax. This does not work well with some, as they are not patient, and cannot adapt to the system's behavior.
So if you move the entire Speech to Text software online, the servers process the audio, using semi-automated means, learn the language of the user and build a specific audio profile online. This will mean some human interaction in the learning process. The main core software along with some operators managing the learning process will make Speech to Text, easy for all.
As the Software is on the servers, it can be updated frequently by the engineers. The dialect database and patterns, can be built during beta and also with every new customer audio profile. The dialect, pronunciation and language templates can easily evolve as the user profile databases are online. Future users may need a learning process which is short, as they have to only match a existing profile template.
Lastly, there is nothing to Install or Uninstall. no Registry to backup. Works from any computer in any country, but only Online. Once in a way, you need to backup your database-setting from your online account on to a CD and keep it very safe.
delabs Notes 2004-2010