FoF Lyric Converter 2.40b (100% success rate)

[raynebc]

I just need some clarification about your request. Rock Band denotes a pitch shift by putting a plus sign (+) after the syllable in question, and UltraStar uses a tilde (~) instead. Do you mean that for the following:

:# # # A
:# # # ~
:# # # pple

you want it to convert as "A pple" instead of "Apple"? Could you provide an excerpt from an UltraStar file as an example of the input, what it converts as and what you want it to convert as?

-- Edit: Mon Feb 15, 2010 4:08 pm --

Updated to version 1.94. Lots of edits, the largest of which being that the "source" MIDI file is now completely optional. If you don't want to create a chart with anything besides the vocals, you can just convert the lyrics to MIDI and it will create a MIDI for you instead of having to copy from an existing MIDI file. The source offset is provided just for granular control in very specific circumstances, such as if you were importing from one MIDI to another MIDI in the same folder. Since the program would normally try to read the offset from song.ini if it was in the folder, it's possible that the MIDI delay isn't the exact same for both charts. You can override the offset in song.ini from being automatically applied to the input file by using the -offset parameter or to the source file by using the -srcoffset parameter. I don't really think there's another circumstance where it would be useful to use unless you had another reason to offset the timing of the source lyric file.

And by request, new logic should now allow whitespace before or after a lyric in an imported UltraStar file to separate it as if it was a complete word. I also put in some new logic to be able to import UltraStar files that use relative timing. I haven't tested these two new UltraStar logic features much, but if they don't work, please let me know. Windows/DOS build is in the first post, Mac build to follow.

Edit:
Latest Linux version is in the first post. Thanks again to trinidude4!

Edit:
Updated to version 1.95. Various fixes and enhancements. I have changed the program to expect vocal pitches in the range of 36 to 84, which is the range documented by Harmonix. Windows/DOS and Linux builds are in the first post.

[raynebc]

Updated to version 1.96
Just like some of the original ripped MIDIs had lyrics incorrectly defined in text events, some of the new RBN MIDIs have game events incorrectly defined in lyric events. I've changed logic to account for this, but now, any lyric that begins with a bracket [ character will be ignored. A new feature is lyricless import, which will allow for import of MIDI, KAR and Vrhythm files where there are no lyrics present. For MIDI/KAR import, this is useful to get pitches when the MIDI author was too lazy to put the lyric text in. For Vrhythm import, it will allow you to just chart the vocal rhythm notes, import and export back to vocal rhythm to create a pitched lyric text file with the correct number of lyric entries, just replace each asterisk with the correct lyric. As an added bonus, if you chart the vrhythm notes so that the first vrhythm note in each line of lyrics is one fret (ie. fret 2) and all other lyrics in that line collectively use a different fret (ie. fret 1), exporting lyricless vrhythm will insert line breaks accordingly. This should take some of the pain out using the vocal rhythm format. Latest DOS/Windows release is in the first post, Mac build to follow. I also tested 140 RBN MIDIs, which all convert after some of the program logic changes.

Edit:
Mac build is now available in the first post. Thanks again to trinidude4.

[SakuyaFM]

I've been getting an overlap lyrics error when converting, as well as a "length of 0" error. What am I doing wrong?

[raynebc]

It really depends a lot on the import file. I have seen some horrendous files before. Feel free to PM me the files involved and the command you're using and I'll try to replicate. Overlapping warnings may be normal for some conversions (ie. from UltraStar).

[SakuyaFM]

The RBN Wiki states that a 1/16th note tail will not show up as a sustain, and must have at least a 1/32 space between notes. Maybe thats part of the problem?

EDIT: I forgot to mention that I dont have the vox chart with said problem anymore as I have given up on it because it doesnt complete with just warnings, it errors out, but I do remember what I did with it. Im just making sure its not going to happen (often).

[raynebc]

It shouldn't be an issue, my converter doesn't deal with that sort of thing.

[SakuyaFM]

"Unexpected end of vocal rhythm notes.
Do not end the pitched lyric file with a line beginning with a hyphen."

But I don't have a hyphen in the end -.-''

EDIT:
New chart is working, but it's got this weird timing problem I can't explain. I know I timed everything right, but sometimes some of the lyrics are not displayed, some parts are off-sync, and some parts get cut off to the next phrase.

[raynebc]

That message should only occur if all vocal rhythm notes from the MIDI have been exhausted and a line break (hyphen) is reached in the pitched lyric file. It doesn't actually verify that there are more lyrics before giving the warning.

[SakuyaFM]

More problems.

Lyrics are synced to the rhythm on EOF, but somehow the game manages to drop lyrics, move notes, and re-set lyrics and phrase positions. What could I be doing wrong?

[raynebc]

FoFiX's vocal stability is very unstable, so there's not much we can do without helping the developers.

[SakuyaFM]

Hmmm, I changed the octaves on the chart, and it seem to have done something. I put it on C1-C3 I lost half the lyrics. I wonder if C2-C4 would be the best? RBN Wiki also said C1-C5 is valid.

[raynebc]

C2 through C6 (pitches 36 through 84) are the valid notes to use.

[SakuyaFM]

Opened my notes.mid file on Reaper, and it's not pretty.

The MIDI text events are off. I don't know if this is an error on my part (I've 10x-checked the syllabication and rhythm on EOF) but a lot of the stuff is off, and the Phrase Markers are off, too.

Gonna recheck again. Text events should help.

[raynebc]

I'm not aware of any problems like that, but more importantly, your lyrics should be stored in Lyric events and not Text events. Other than that, you really haven't given me enough information to help you. For better results, I'd need input files and audio so I can find out if I need to correct something or if you need to correct something. If you don't want to go that far yet, then at least send me a debug log (use the -debug parameter and redirect the program output to a file).

[raynebc]

At long last, the next major release (2.0). There are an extreme number of improvements so the quick summary is two new export formats, four new command line parameters and lots of fixes. Also, this program is now compiled into EOF, supplying the guts behind its new Import Lyrics feature. No longer will people need to use a command line program to merge lyrics into their chart, they can import directly from EOF and immediately test the lyrics in their chart, pitches and all. The Windows/DOS binary is in the first post, Mac binary to follow.

Edit:
I found I posted the wrong Windows executable, so for the one person that downloaded it, you may find that it doesn't work. I put the correct Windows release in the first post.

I hadn't done a stress test in a while, so I am running a script to test importing all 1190 or so RB1, RB2, RBB, RBN and DLC MIDIs and exporting them to all supported formats. The detection logic showed that 22 of the MIDIs didn't have vocals, so for each of the MIDIs that do have lyrics, I'm expecting one file for each output format (script, VL, MIDI, UltraStar, LRC, Extended LRC, Vocal Rhythm MIDI, Pitched Lyric file, KAR, Soft Karaoke) to be exported by the converter, for a grand total of around 11680 lyric files created in total. The lyric conversions are likely to take several hours, so I'll post the results some time tomorrow if it finishes.

Edit 2:
After removing some duplicate MIDIs from my giant test Rock Band folder, I was left with 1176 MIDIs, 19 of which have no vocals. 41 wouldn't import for some reason, bumping the success rate down to 96.5%, which is still pretty impressive considering I just tested against several hundred new files. I'll make resolving those import issues my first priority and release another update.