|
|
View previous topic :: View next topic |
Author |
Message |
ajanata
Joined: 07 Jul 2007 Posts: 1167 Location: South Bay Area, CA
|
Posted: Fri Mar 28, 2008 8:30 am Post subject: |
|
|
I don't think that's the case, on THTF, they tend to use the same character throughout a single phrase. I made a quick early alpha of THTF to demonstrate what I'm talking about. (Ignore the fact that the pitch lines make little sense in non-talky parts. I'm still working on that.) _________________
|
|
Back to top |
|
|
GuestWednesday
Joined: 01 Jan 2008 Posts: 314 Location: UK
|
Posted: Fri Mar 28, 2008 11:31 am Post subject: |
|
|
Sorry, didn't desribe THTF very well. As ajanata's pic shows, just two out of the "Will you stay down on your knees?" are ^ at the end, the rest is #... and this is where I usually lose points on the PS2 version.
Most phrases just use one of the symbols. The most extreme example I've seen is in Highway Star:Code: | ^ # * ^ ^ # *
I love it and I need it |
There's also Enter Sandman:Code: | ^ ^ # ^ # ^ #
If I die be-fore I wake | and Code: | ^ ^ # # # ^ ^
Oh my Suf-fra-gette Cit-y |
and in SabotageCode: | # * # * * # * # *
I can't stand it I know you planned it | but most of the song is one type per phrase, including later when it'sCode: | # # # # # # # #
Can't stand it I know you planned it |
|
|
Back to top |
|
|
NumberCruncher
Joined: 07 Dec 2007 Posts: 93
|
Posted: Fri Mar 28, 2008 1:41 pm Post subject: |
|
|
The alternating parts in THTF don't really fit this pattern, but from the examples GuestWednesday posted, it looks kind of like ^ parts are more on the voiceless end of the spectrum, while # parts are more voiced. As for *, and this is just a wild guess, it might indicate more of an emphasis on timing. |
|
Back to top |
|
|
LocalH
Joined: 30 Oct 2006 Posts: 1400 Location: MiloHax
|
Posted: Sat Apr 05, 2008 4:53 pm Post subject: |
|
|
Ok, I'm going to post what I've found by looking at the PS2 dta files (specifically, config/midi_verify.dta). Some of this will duplicate what has already been posted but this is straight from the dta so it should be accurate. Descriptions are mostly verbatim copypasta from dta comments ("midi:" refers to note numbers in that track, "allowed" and "required" I assume refer to text events:
Allowed tracks:
'PART DRUMS'
'PART BASS'
'PART GUITAR'
'PART VOCALS'
EVENTS
VENUE
BEAT
Track events
---------------
BEAT:
midi: 12, 13 - tempo beat track
EVENTS:
midi: 24-26 - practice mode beat box
required: music_start, music_end, end
allowed: coda, section, crowd_realtime, crowd_mellow, crowd_normal, crowd_intense
VENUE:
midi: 24-26 - click track
midi: 37-40 - spotlight focuses
midi: 48-50 - lighting keyframes
midi: 60 - camera cuts
midi: 61-64 - camera char focus
midi: 70-73 - camera shot type
midi: 85-87 - band members sing-a-long
midi: 96-110 - post proc notes
allowed: chorus, verse, lighting, do_directed_cut, do_optional_cut, bonusfx, bonusfx_optional, FogOn, FogOff, StrobeOff, Strobe6, Strobe8, Strobe10, Strobe12
'PART GUITAR':
allowed: MUSICIAN_COMMON_EVENTS, play_solo, map
midi: 12-15 - h2h camera cuts and focus notes
midi: 40-59 - fret animation
midi: 60-62 - easy gems
midi: 67 - easy solo marking
midi: 72-75 - medium gems
midi: 79 - medium solo marking
midi: 84-88 - hard gems
midi: 91 - hard solo marking
midi: 96-100 - expert gems
midi: 103 - expert solo marking
midi: 105-106 - multiplayer phrases
midi: 116 - spotlight phrases
midi: 120-124 - big rock ending
'PART BASS':
allowed: MUSICIAN_COMMON_EVENTS, play_solo, map
midi: 12-15 - h2h camera cuts and focus notes
midi: 40-59 - fret animation
midi: 60-62 - easy gems
midi: 72-75 - medium gems
midi: 84-88 - hard gems
midi: 96-100 - expert gems
midi: 105-106 - multiplayer phrases
midi: 116 - spotlight phrases
midi: 120-124 - big rock ending
'PART DRUMS':
allowed: mix
allowed: ride_side_true - ok to use side swipe on ride cym when hits are slow enough
allowed: ride_side_false - never use side swipe on ride cym
allowed: MUSICIAN_COMMON_EVENTS
midi: 12-15 - h2h camera cuts and focus notes
midi: 24-27, 30-31, 34-42, 46-51 - animation
midi: 60-64 - easy gems
midi: 72-76 - medium gems
midi: 84-88 - hard gems
midi: 96-100 - expert gems
midi: 105-106 - multiplayer phrases
midi: 110-112 - tom gems
midi: 116 - spotlight phrases
midi: 120-124 - drum fills
'PART VOCALS':
allowed: tambourine_start - switches to tamb anim cut on next camera cut
allowed: tambourine_end - switches back to normal anim groups
allowed: cowbell_start, cowbell_end, clap_start, clap_end, MUSICIAN_COMMON_EVENTS
midi: 12-15 - h2h camera cuts and focus notes
midi: 40-84 - transcribed vocal melody
midi: 96-97 - percussion gems (tambourine, cowbell, etc)
midi: 103 - percussion phrases
midi: 105-106 - multiplayer phrases
midi: 116 - spotlight phrases _________________
|
|
Back to top |
|
|
httenrai
Joined: 14 Apr 2007 Posts: 16
|
Posted: Wed Apr 16, 2008 11:35 pm Post subject: |
|
|
I think I'm on to something here. Let me see if I can explain this correctly:
Code: | ^ # * ^ ^ # *
I love it and I need it
I luh ih ah I nuh ih
^ ^ # # # ^ ^
Oh my Suf-fra-gette Cit-y
Oh muh Suh Fruh Jeh Zhih-Teh |
If I'm reading this right, the game is detecting three kinds of phenomes. I think the mic is looking for different sounds to gauge whether the player is singing the words properly. I'm no expert on phenomics, so I couldn't classify them, but just look at the first sound you make on each one. ^ might be an easily detectable sustained sound, # might be detecting abrasive sounds like hisses and scratches, and * might be looking for short, abbreviated sounds that may not resolve with an audible consonant, maybe.
I wonder if you could get the pie full by doing something like this:
Code: | ^ ^ # # # ^ ^
Oh my Suf-fra-gette Cit-y
Uh uh TSuh TSuh TSuh Uh-Uh |
What do you think? _________________
Current goals:
Change goals [x]
GS GGandHT [x]
Stop abbreviating things [ ]
Forget about Jordan [ ]
Get divorced [Thank deity of choice]
Beat Endless Setlist [ ]
Not kill someone who tells me to practice [ ]
Lyrics and pitches for all songs on disc |
|
Back to top |
|
|
GuestWednesday
Joined: 01 Jan 2008 Posts: 314 Location: UK
|
Posted: Thu Apr 17, 2008 12:01 am Post subject: |
|
|
httenrai wrote: | I think I'm on to something here. |
Those were the most extreme examples I could find of varying talky types within the same phrase. Often, entire songs only use # for talky parts Even Sabotage is almost entirely # except for a few phrases.Code: | # # # # # # # # #
Schem-ing on a thing that's a mi-rage
# # # # # # # # # #
I'm trying to tell you now it's sab-o-tage |
The ratio for talky syllables on-disc is about 75% #, 12% ^, 13% *. You can see what all on-disc songs look like in this thread http://rockband.scorehero.com/forum/viewtopic.php?t=7164
I'm still pretty sure # is "normal talky" (just because it's so common) and * is meant to be shouting/screaming (based on where nearly everywhere it's used) and I've not been convinced by any explanation of ^ (usually seems similar to shout/scream).
Does anyone know what the other files for each song are for? Are they the same as GH or KR? Could any of them be data for phoneme detection? Every song on the PS2 disc, whether or not they have talky parts, has .xvocab .voc (which sound like they should be related to vocals... lip sync? ) .pan .usr .vnn as well as .mid... and the definite audio/video files. Is talky part detection done just based on the MIDI lyrics and suffix modifier character or does it compare it to the vocal master track or other files? |
|
Back to top |
|
|
Kawigi
Joined: 27 Feb 2008 Posts: 2879 Location: Redmond, WA
|
Posted: Thu Apr 17, 2008 12:25 am Post subject: |
|
|
I think there's got to be some kind of real phoneme detector in there, possibly using one of those files, even if its effectiveness seems to be overshadowed by more general things (like these suffix characters and rhythm). But I'd be pretty convinced that one of those files has to do with lipsync - the characters animated in the game clearly have some idea how one would move their mouth to sing the lyrics, and that information definitely isn't anywhere in the midis. |
|
Back to top |
|
|
httenrai
Joined: 14 Apr 2007 Posts: 16
|
Posted: Thu Apr 17, 2008 12:33 am Post subject: |
|
|
Well, in the examples I had, it was shifty at best, but then you pulled out Sabotage. So, either they specifically altered the detection values for Sabotage to be more lenient (as I find them) or I'm just up crazy creek without a paddle. :P
I thought I was on to something, though. I might still be if my former assertion above holds true.
EDIT: Nope. I looked at it. My theory is as sound as a Brawny paper towel with a hole already punched through it. Crap. _________________
Current goals:
Change goals [x]
GS GGandHT [x]
Stop abbreviating things [ ]
Forget about Jordan [ ]
Get divorced [Thank deity of choice]
Beat Endless Setlist [ ]
Not kill someone who tells me to practice [ ]
Lyrics and pitches for all songs on disc |
|
Back to top |
|
|
httenrai
Joined: 14 Apr 2007 Posts: 16
|
Posted: Thu Apr 17, 2008 2:30 am Post subject: |
|
|
Ok, here's an off-the-wall theory: Maybe they have something to do with the singer's eyebrow positions?
I can only think of three states: Level (straight), furrowed (down), and enlightened (up).
Maybe double check the lyrics with someone's eyebrows... O_o
/From downtown....wild shot.... _________________
Current goals:
Change goals [x]
GS GGandHT [x]
Stop abbreviating things [ ]
Forget about Jordan [ ]
Get divorced [Thank deity of choice]
Beat Endless Setlist [ ]
Not kill someone who tells me to practice [ ]
Lyrics and pitches for all songs on disc |
|
Back to top |
|
|
Qwertyuiopia
Joined: 17 Feb 2008 Posts: 53 Location: under your bed
|
Posted: Sat Apr 26, 2008 9:25 pm Post subject: Rock Band -> GH2 |
|
|
I actually extracted the midi for Enter Sandman (from Rock Band), opened the mp3 (I already had it on my computer) in FeedBack, copied the midi into FeedBack, exported it (as a .chart), replaced a bonus song in Ghex with the new chart, burned it to a dvd-r and it actually worked! I even got the offset right. It works prefectly good!! I don't know if someone already did that or not... just wondering if it is of any interest. By the way this was for Guitar Hero 2. _________________
|
|
Back to top |
|
|
laconic
Joined: 24 Jan 2008 Posts: 243 Location: Saint Peters, MO
|
Posted: Sun Apr 27, 2008 2:37 am Post subject: |
|
|
xvocab is related to Fonix VoiceIn SDK, so is usr. See here, then download the demo. Rock Band uses the Fonix VoiceIn SDK for it's voice recognition, see the last couple pages of the Rock Band manual for proof.
I was actually looking into this a couple months ago, but I couldn't find anything online about programming with the SDK. Although, I didn't look real hard, but a quick look now turns up nothing.
Maybe I should try looking up info on the Unreal 3 runtime... |
|
Back to top |
|
|
TNTX
Joined: 26 Oct 2006 Posts: 1069 Location: Grand Island, NE
|
Posted: Tue May 27, 2008 3:30 am Post subject: |
|
|
For Anvil Users: To get the correct note set, once in Piano Roll mode, click Add Sounds, and select C3, set to 128 notes. This SHOULD get you all the way up through overdrive (G5). _________________
|
|
Back to top |
|
|
paul4806
Joined: 11 Mar 2008 Posts: 1
|
Posted: Sun Jun 15, 2008 3:38 pm Post subject: |
|
|
Incase anyone is interested, rockband doesnt look for words or phrases so much, it merely looks for the right note at the time it is expecting one. I have made a device that plays the notes from the midi files at the right times and it can pass the harder songs on expert with 97% ish. |
|
Back to top |
|
|
TNTX
Joined: 26 Oct 2006 Posts: 1069 Location: Grand Island, NE
|
Posted: Sun Jun 15, 2008 6:23 pm Post subject: |
|
|
paul4806 wrote: | Incase anyone is interested, rockband doesnt look for words or phrases so much, it merely looks for the right note at the time it is expecting one. I have made a device that plays the notes from the midi files at the right times and it can pass the harder songs on expert with 97% ish. |
I'm guessing the only thing it misses are talkies for the most part? Just a solid C5 might nab them, as PS2 Talkies aren't as strict on it. _________________
Last edited by TNTX on Mon Jul 07, 2008 11:11 pm; edited 1 time in total |
|
Back to top |
|
|
TNTX
Joined: 26 Oct 2006 Posts: 1069 Location: Grand Island, NE
|
Posted: Mon Jul 07, 2008 11:04 pm Post subject: |
|
|
JetsterDajet wrote: | How many custom chart creators have the capacity to transpose and notate music by ear? The sad truth? Only a handful, which will lead to a relatively small pool of RB customs with vocal tracks.
Sure, some people might be able to scrape by with an electronic tuner, but it won't be the same. |
Honestly, it's not even that hard. All I do is get a midi of the song (5 minutes in google for most songs), stick it in KRMaker with the .mp3, make sure it's lined up (or, stays out of sync at the same spot through the whole song. If it's a half second ahead at the start, just make sure it's a half second ahead at the end). From there, insert lyrics, and line up with guitar and drum parts with a midi editor. _________________
|
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
Copyright © 2006-2024 ScoreHero, LLC
|
Powered by phpBB
|