Well, now we're even - because I don't understand the above program.
I would throw this "fish" back, and go study the last program before this. That one was the way to go.
This one is not a "keeper".
You need an array of strings that will associate the word "two" (for instance), with the word[][] array row 2. Like the last program had.
Then, when you are scanning through a number, and you come to the digit 2, you know where to find the right word - in word[2], which equals, LOOKY THERE, IT'S A MIRACLE!!, "two".
There are other ways to do this, but they pretty much follow this same scheme or organization. Word[][] would be a 2 dimension char array with entries like: "zero", "one", "two", "three", "four", etc., which would correspond with word[0], word[1], word[2], word[3], word[4], in rows. (So word[0] is a whole row, not just one char like word[0][0], would be.)
Let's look at word[0]:
It has 6 subscripts (spaces), because "seven" has 5 letters, and then we need one char for the end of string char: '\0', so it's elevated to string status, not just a few char's.
So word[0][0] == 'z';
word[0][1] == 'e';
word[0][2] == 'r';
word[0][3] == 'o';
word[0][4] == '\0';
But - very sweetly, word[0] == "zero", (yes, the whole string!).
That's what you need - that's what you need to study - not this last misguided effort.
And if I were asking for help on an English help forum, I'd have the program in English, or some sub-titles in English. Obviously, you speak it well enough to do that.
P.S.
Ignore comma's. You can't rely on numbers always having them (2,000 does, but 2000 doesn't), and you don't need to do any special processing for them. The word to remember for comma's is "continue".