I noticed dictation only works if you have internet connection, and with only a select few languages (English, Japanese, but not Spanish when I tried).
My guess is a recorded speech is digitized, compressed, sent to some servers (iCloud? or Wolfram Alpha?) for transcribing, then sent back and pasted to document (very impressive).
It also seems to analyze the sentence to help find the correct word, instead of simply analyzing the audio.
It is when trying to speak in a combination of numbers and letters, however, it fails.
For example (take from another discussion topic):
Would like to know more about how is works, and suggestions on how to input speech in no grammar entry, that is pure Alpha-Numeric strings.
My guess is a recorded speech is digitized, compressed, sent to some servers (iCloud? or Wolfram Alpha?) for transcribing, then sent back and pasted to document (very impressive).
It also seems to analyze the sentence to help find the correct word, instead of simply analyzing the audio.
It is when trying to speak in a combination of numbers and letters, however, it fails.
For example (take from another discussion topic):
Any data entry into the iPad that switches between letters and numbers is somewhat cumbersome.
Entering "A11 DEF V111 GHI V222 JKLMN Q9A" could possible be easier to speak than type.
In non-rigorous testing, though, I get a weird assortment of results, even though I'm speaking one letter or number at a time. For example, sometimes my home airport, I19, shows up as "I19", "I 19", "EYE19", etc. I think the iPad doesn't know what to do with it since it's not a word.
Would like to know more about how is works, and suggestions on how to input speech in no grammar entry, that is pure Alpha-Numeric strings.