On a random subject, how big did you make the ROM? IIRC, FFVI is already 32Mbits (4Mb), which is the theoretical max for a SNES game...

To Ryusui: I downloaded a 3,072 kb rom. Now it's 4,096. So, 4Mb I'd say? Please don't tell me I need to download a different rom lol

To Vehek: The dialog goes into bank $CE and ends there, but starts in bank $CD at CD0000. I used Thingy32 (ff3us.tbl and ff3btl.tbl) to upload tables and that where it started.

Your peers have reported this post as being problematic and interfering with their ability to enjoy the forums and healthy community atmosphere. The moderation staff has evaluated these reports and taken appropriate moderation action and warning involved individuals. Please take severe disputes and non friendly or respectful discussion into private conversation and be considerate of your fellow community members.

Additional moderation action has been taken. We understand that you may object to the previous moderator interjection. However, when your fellow community members tell us there is a problem, we must listen. We are doing the job your peers have asked for. Please respect moderator interjection on behalf of your fellow community members. The community feels strongly about upholding a friendly and respectful atmosphere. Feel free to contact us to discuss the matter for further clarification. If community members continue to report a problem here, more severe moderation action will be taken. For the benefit and consideration of all, please discontinue the current line of problematic discussion.

If you're using Geiger's snes9x debugger, this should be a snap.  You know what the first string says, right?  Then put a read breakpoint at CD0000, and get to the part of the game where that string comes up.  The break will occur when it's accessed.  If you see:
someplace: lda $CD0000,x
then put the new 24 bit address of your text at someplace+1.  Otherwise, the breakpoint came just a few instructions too late to tell you where the pointer came from.  Still, do pay attention to that instruction's address mode and operand.  When the break occurs, hit 'Step out'.  Jot down the address of the next instruction that you see.  Subtract one from that address and set it as an execution breakpoint.  Do whatever you did to get the text to come up again and the new breakpoint should come up first.  It will be a jsr or jsl instruction.  Hit 'Step into' to continue into the subroutine, and that will basically spell it out for you.

Remember how you paid attention to the address mode and operand of the instruction that LDAed the first byte of the string?  That's because you need that information to know what you're looking for earlier in the subroutine.
If the first breakpoint turned up any of the following address modes:
(something),y
something,y
something,x
Then you're looking for a set of push (PH* or PEA) and pull (PLB) instructions; the register that is pushed last was carrying the bank part of the pointer, and you can trace it back to some place in the rom from there.

Otherwise, you saw:
[something]
or
[something],y
which means that you are looking for an earlier store (STA, STX, or STY) to something+2.  Again, the stored register is carrying the bank, trace it back to the ROM from there.

If that doesn't help at all, look for a doc on basic 65x assembly.  There are plenty, and they should make that make some sense.

ALL of it?  Probably not.  Regarding the stuff I wrote, the general technique of using a debugger to find out where something happens transcends processors and architectures, but I can't see how a lot of the stuff about address modes would apply to the NES.  I don't know a lot about it, but I'm pretty sure there's no way to fetch data from an offset into an array through a far pointer in RAM in a single instruction.

But can you be a little more specific?

No doubt lol. Decided to check out this forum to see what was going on. Gonna be a long list of thanks in the end credits of this mother.