And no, keep the original data so that CRC matches (that's what the header skip option in GENH is for).
Here's how it works, so far as testing on 2 seperate files:
That bit value here in my previous post:
"F8 33 00 00 - I don't know again"
This translates to 13304 decimal. Multiplay by 16 to get 212864 bytes, then add the header byte count, and you get the loop start point to enter into GENH.
Strangely, after looking at them again, converting the value you get after multiplying by 16 back into hex, I noticed that the loop point IS exactly what it states, except it's one bit off:
F8 33 00 00
After multiplying and converting, you get:
33F80
or what it seems like it should say in the header: 80 3F 03 00, or maybe -0 F8 33 00
I'm going to go ahead and make GENH files out of these, label, and upload for you guys. Now that I've actually found the loop point, it'll be a snap to do.
As for manakoATs files, did he delete all the redundant blank data that literally takes up megabytes of space (no header or any other info, just spacer), and get the right loop points, because last time I talked with him, he had it off by a channel so that you'd actually have one channel's loop start cut off, which can cause there to be a very slight pop in a single channel at the loop point.
I've already gotten more than 3/4 of the way done on mine last I got to work on it, and have my own program to find exact sample count at the loops, so I'm still gonna finish my own set. If I get ahold of the the completed set, I'll compare and see how that goes.
edited 6:22 PM EDT March 23, 2009
Changed a transposed hex number...
edited 6:35 PM EDT March 23, 2009
In the actual single chunk of a file all of the music comes from for WA3, there are gaps between the songs that are megabytes in size and account for sometimes up to 30 seconds of silence when played.
What's more, I found a bunch of songs that have pieces of the song scattered around in different places in the file and aren't connected, but they are SUPOSSED to be a single song. I'm using the soundtrack to figure out where the pieces go.
A third thing I had to go through was that, with some of those songs that are in pieces, there are duplicates of some of the pieces that may or may not be the same. They may actually be different pieces that are to make 2 slightly different versions of the same song, like one that continuously loops at a piece, and another that plays straight through, or 2 that have a slightly different piece.
There were tons of these pieces of songs that are not individual songs on their own, but programs split them because it appears to be, and even when I manually went through and split the file down, I couldn't tell, either, until I listened to them.
One final thing I was doing that I had to go recheck was that some songs have a trailing ending that appears as if the song data had ended, but in fact, I had accidentally cut off some extra data, and had to go back and figure out where the song actually turned to being complete silence (nothing but 00).
For MMX7:
I'm doing some experimenting with the loop start before I start making them. After messing around, I'm wondering if the loop start in these headers takes into account the header size at all, because weather I have it on or not doesn't seem to make a difference except that it just barely seems to sound as if the song loops a tiny bit ahead with the header size added. It just feels like the beat is a tiny bit off at that point or something.
Hell, with the header being for the wrong file type, the loop start seeming to be moved a single bit throwing it off (maybe it's already converted and multiplied by 16 in hardware), I wouldn't be surprised if that was figured that way.
Edit:
I'm still experimenting with whether the header is counted or not, but have a question: Since I've found the loop start value and how to get it, is it now possible to have vmgstream read and play this, loop and all, without making these into genh files, ignoring the file type and that other value you mentioned vgms reading? It'd save a lot of trouble, too.
Whatever would be simplest to do.
edited 8:03 PM EDT March 23, 2009
edited 5:56 AM EDT March 24, 2009
WA3 stuff, what's wrong is wrong, plain and simple. I spent way more than enough time manually going through the music code, breaking it down, making the looped genh files, and carefully checking them to tell how it's read. Last time I talked with manakoAT, he wasn't planning on working on the music anytime soon, said he wasn't even close to being done, and said it was fine that I was doing it instead.
He also had one thing wrong when I was talking to him about the loop starting point, which was, instead of taking into account both channels for the interleave (2048 byte increments to get the data for BOTH channels for interleave of 1024), he had figured at the wrong spot (only in 1024 byte increments), needing it to be moved ahead by another 1024 bytes to get the sound data for the other channel at the loop point. I figured this out myself AFTER I had last talked with him about the calculations and stuff to find the loop value.
MMX7 stuff, depending on whatever you're referring to, I have no idea how the creators put these together, so I don't have a clue how the loop start point is read by the game. Did they count the header or not? It isn't counted in the file length value. Did they get that start value before slapping the incorrect header on, or after? Noone knows but them.
VGMstream/MMX7 stuff, I don't get what's wrong here, I asked if this game can have some exception added into vgmstream telling it to ignore what's wrong in the header, read the good data, and play. I know there are other games that already HAVE this in vgmstream, because Gradius V was one I told hcs about, which was playing at super speed, exactly like this does, and he had to change something so those would play properly.
nonsense, you asked me tons of questions and quit the channel, and btw it was all the ripped, you can download it on TPB or wherever and look at the files, they're using all the 1024 byte shit...
You can trace/hack/whatever down what you want, i don't know why you make it, there's a simple table in the main executable where all offsets/loopstarts/loopends are stored in, i told it to you thousand + 1 times...
and now, please quit this nonsense here, rip your set and be a hero, i don't care about it!!!
edited 6:28 PM EDT March 24, 2009