Suppose you say "Windows" then

1) Imagine the audio graph as |-----Windows---|  for sample processing resolution  "COMPLETE" then during Recording for Addition to prepare a basic Vocabulary SAY file you will hear a complete word ,so,  store it giving the same name "Windows". Prepare five or more such files with different pronounces of Windows in each file. Experiment with accuracies  (their is no need  to enable Artificial Intelligence or start language editor in this case ). Note that Windows is a mixure a lot of frequencies so this method is technically inaccurate.In case you get bad results especially when you record more than 10 words in each file, try the second method given below which is the most recommended one.

In this case the Golden _Chara will give you the final result. Please enter  all the Windows in every file without a spelling mistake and of similar characters, do not expect windows to match WINdows.

2)  Imagine the audio graph  |--------~win~~doo~~ws~~------| for sample processing   resolution to be 4000 for a frequency setting of 22.5Khz. i.e if your Windows is 12000 bytes length word,  it will   get divided into 12000/4000 = 3 times i.e  1)eein 2) oo 3) ss. Store these in your vocabulary file and prepare five such files (Five times recording) and do this for all your words. Do not ADD again and again "oo" if your second word happens to be "Doors" (Confused ? recall your first word here is Windows).

See how strings are reproduced by the Golden_Chara, enable Artificial Intelligence, see them 10 more times to finalize a protocol for all your words(see how to prepare them on the Language Editor screen ).

Please enter  all the 1)eein 2) oo 3) ss in every file without a spelling mistake and of similar characters, do not expect 'oo' to match 'o'.