[maemo-developers] One question about speech2text poor performance

[From: Marco Solari]

Hi!
I have a question about the voice recognition problem on the Nokia platform,
hoping it's not too off-topic...
I see on this platform some voice recognition tool is available, but it's
extremely resource demanding, slow and performs with quite a low success
percentage... I'm asking myself how comes today quite every low-end phone is
capable to recognize uttered names by a not-trained user, with 90% success
rate (I would estimate); I believe Nokia platform sports quite higher
hardware (CPU/RAM) than thos pohones...
The question is: is the task of recognizing a small set of utterances
basically different from the task of recognizing spoken words out of a
bigger dictionary (I limit the problem to single words, not considering the
spoken speech...), or the commercial phones name-recognition software is
really more advanced than open source tools we use ?
Thanks in advance for your attention.

Marco Solari