GPA synced XML "drags" behind the beat when imported into EoF

[Billkwando]

I've been having this problem for a while, so I upgraded from a 2sh version of GPA to 3.42 (the newest version I could get w/o paying again, plus the newest version's "auto sync" feature sounds limiting to me, because I'm afraid of not getting a good sync and being unable to adjust it manually (not sure if this is the case). I set a system restore point so I can downgrade to the old version if needed.

 

Anyway, I have 2 or 3 songs that when I import them to EOF and turn on the metrognome to test, the beats seem very obviously off, across the whole chart. 

 

Could someone take a look and see what it is I might be doing wrong? As some of you may know, I'm pretty experienced at this. I'm planning for the next 6 L'Arc~en~Ciel song CDLCs, and I want to do a sweet song in celebration of my 50th L'Arc CDLC (I've made around 45 or 46 now), and sometimes it feels like I'm never gonna get there, because all of the songs I want to do won't import right.

 

Here's an example if someone would test:

 

Shinjitsu to Gensou to (GPA Library export):

 

https://mega.nz/#!Vo0HCJDa!8AbcYLMvZNYowChuCGvHK8S5Rb4XYnByJa5foW3nGYA

 

Here's the EOF folder with the audio and stuff:

 

https://mega.nz/#!1x1HERpa!tggLk_zafPyrxf_FjDl86EjsuF5YUsGJFEiZfeo-CmY

 

 

In the past I've fixed this by dragging the beat map to match the music, and then pasting the notes back in, but obviously that's not ideal.

 

Thanks for looking!

 

Bill

[firekorn]

Well that's because the two audio file aren't starting at the same time :

 

 

The top one is the one used by EOF and the one at the bottom is the one that was used for GPA sync.

 

Note that you can quickly fix it by simply draging the first beat in EOF (which will move everything together).

[Billkwando]

Well that's because the two audio file aren't starting at the same time :

 

 

The top one is the one used by EOF and the one at the bottom is the one that was used for GPA sync.

 

Note that you can quickly fix it by simply draging the first beat in EOF (which will move everything together).

 

 

I am a spectacular new species of dumbass, worthy of study by anthropologists.

 

I do indeed have a whole other folder of L'Arc mp3 (down from about 5 of them) and GPA was pointing to the wrong one.

 

 

About the dragging, do I need to do select all or something? I have to import a track to get the beat map to move to the right-(ish) place, so there's no opportunity to do without the notes on there.....also, won't it try to move the messed up beats back every time I try to import a new instrument to my difficulty tabs? What about the bass track?

[Alex360]

@@Billkwando another option would be import gpa sync before adding leading silence to the track and enable adjust notes for bpm markers then do a leading silence thing ;)

[Billkwando]

@@Billkwando another option would be import gpa sync before adding leading silence to the track and enable adjust notes for bpm markers then do a leading silence thing ;)

 

I always import before adding leading silence, and I'm not even adding it to this one because of the long piano intro (guitar/bass comes in later). Where is "adjust notes for BPM markers" located? Does that make it drag the notes with the beats?

[firekorn]

@@Billkwando just select and drag the very first beat and it'll move everything along.

[Billkwando]

@@Billkwando just select and drag the very first beat and it'll move everything along.

Except it doesn't, when I try it:

 

 

You do mean to grab the first grid line of the first measure and drag it to match the waveform start, right?

 

In my case, the notes were left behind.

[firekorn]

The note should move along with the beatmap unless you disable the note auto adjust in the preferences but in the worst case scenario a copy/paste should allow you to easily put the tab back into place.

[Billkwando]

The note should move along with the beatmap unless you disable the note auto adjust in the preferences but in the worst case scenario a copy/paste should allow you to easily put the tab back into place.

 

OK, it turns out that that option was disabled, not sure why. I figured out that you do have to import all your instruments first, otherwise it moves the chart back to where it started, but the notes are left behind.

 

I'm still having the sync problem with GPA to EOF, and I did double check and make sure they were using the same source mp3. I actually deleted the EOF folder for the song just to make sure I was using the right file, and I checked the file path on GPA to make sure it was pointing to the same folder as EOF.

 

I even tried creating the EOF project by browsing to the mp3 in the GPA export itself, and it still was off.

 

Here's the file if someone wouldn't mind taking a look. I'm stumped for why it's happening.

 

https://mega.nz/#!hxsTFQSJ!o2YLagljOXMJBej8KfU0ZUucdv99TvQM8qlMTvJpw7o

 

 

BTW the dragging thing works awesome as long as I have everything imported first. :D

[raynebc]

When I import that GPA file and use the provided MP3 file, the waveform graph shows that the sync is off by 50+ ms as early as the second beat marker. If you open the EOF log after importing the GPA XML file, it lists all of the sync points defined in the XML file, such as:

Sync point #0: Pos: 634ms Measure: 1.000000

Sync point #1: Pos: 6085ms Measure: 3.000000

...

Measure 1 is where the first beat marker is positioned and if you compare with the waveform graph, it's 50+ ms out of sync with the spike in the graph that represents the snare. If you go to the second sync point (measure 3), the graph shows that the timing is off by about 60 ms. EOF appears to be importing the file correctly, but the syncing done in Go PlayAlong needs more work. You can alternatively fix the sync in EOF, but it's up to you to decide which (fixing in GPA or fixing in EOF) is less work.

[firekorn]

@@raynebc Except that the marker aren't actually 50ms off compare to the original file :

 

https://imgur.com/L0r0iEQ

 

It's about 18ms give or take, quite far from the 50+ms that seems to happen when you go into EOF :

 

https://imgur.com/SWNbzZF

 

So i decided to check where things go off by comparing the original MP3 with what EOF create :

 

https://imgur.com/XeTCsJS

 

The first is the original MP3, the second is the WAV file create by EOF and the last one is the guitar OGG created by EOF and here i'm stumped...

 

When the MP3 is converted to WAV, the leading silence is shortened so i wanted to see if it was EOF specific just to be sure by doing the conversion with another software and i'm not sure if i should have expected that result or not :

 

https://imgur.com/0zNNhXC

 

So here you have the original MP3 against the WAV made by another software and the result is the same as above.

 

That MP3 for an unknown reason when converted to WAV have less leading silence which of course makes the sync go off...

 

Not sure what's the take from this but it's clearly something i wouldn't have expected from a file format conversion.

[raynebc]

When I have Audacity convert the MP3 to OGG, it does reflect this difference. After further testing, this appears to be a bug with LAME (mp3 encoding/decoding utility). The intermediate WAV file created when decoding the original MP3 file reflects the missing ~50 ms of audio at the beginning. I'm going to have to see if this is something I can do anything about (ie. finding a non-broken version of LAME).

 

Though this LAME bug is the biggest problem, there are still some other, lesser sync problems to correct.

 

Edit: Actually this seems to be something odd with the MP3 itself, ie:

https://sourceforge.net/p/bonkenc/bugs/88/

I used Audacity to convert the original MP3 to WAV, then used LAME to encode to MP3, and then gave that MP3 to EOF. It converted that to OGG and there was no missing audio at the beginning. This appears to be a limitation with LAME 3.99.5 (the version that Windows EOF bundles with).

 

Edit 2: I opened a bug report with the LAME project to see if they will pursue this:

https://sourceforge.net/p/lame/bugs/486/

[Billkwando]

Thanks guys! It's nice to know I'm not crazy.....I've been making these things a long time! ;)

 

I knew it couldn't be simple user error, because when I drag the beat to match the wave graph, it syncs perfectly.

 

I was telling @firekorn the first thing I do when I import a GP tab is listen to it in EOF with the metrognome turned on, to make sure it's lined up and the beat matches. That's how I spotted the problem.

 

If nothing else, at least you guys have heard L'Arc~en~Ciel now, and one of their really good songs, at that! Lots of tasty guitar. ;)

[raynebc]

They are definitely good and are already one of my favorite Jpop bands. "My Heart Draws a Dream" is probably my favorite song of theirs.

[Billkwando]

They are definitely good and are already one of my favorite Jpop bands. "My Heart Draws a Dream" is probably my favorite song of theirs.

 

Oh neat, I didn't know that! :)

 

My Heart Draws a Dream is on my to do list! Feel free to offer feedback on any of my L'Arc CDLCs. I'm often updating them, so if you think one needs something, I'm happy to revisit them.

 

I even have 2 Ken signature guitars that I use when I'm setting up the tones, one for early stuff (Fernandes LA-85 KK), and a LD-95 KK for everything else (with some overlap between....driver's high needs 24 frets) LOL. Cos I'm a goof.

 

I'd gotten about 30 CDLCs in, and when I found out Ken guitars actually existed, I was like "Dammit, I'm gettin' me some L'Arc guitars, I've earned it!" LOL. Fortunately, my wife plays guitar and likes L'Arc, and thus didn't grumble too much.

[Alex360]

@@Billkwando, what a lucky bastard you are, huh :D
I already jelious for this a bit ;)
LMAO

I should get explorer from '86 and setup it for baritone yeah that would be nice to have

[raynebc]

So far, LAME support appears to believe that the application and Audacity are misinterpreting the MP3 file by treating the "encoder delay" and "LAME tag" as audio data when they should be discarded, adding around 39ms of extra audio padding at the beginning of the file. Looks like this is the most likely explanation for why this happened: Some exotic MP3 codec metadata that isn't reliably supported between different programs.

[Billkwando]

So far, LAME support appears to believe that the application and Audacity are misinterpreting the MP3 file by treating the "encoder delay" and "LAME tag" as audio data when they should be discarded, adding around 39ms of extra audio padding at the beginning of the file. Looks like this is the most likely explanation for why this happened: Some exotic MP3 codec metadata that isn't reliably supported between different programs.

 

Is this something that can be fixed, or are you still looking into that?

[raynebc]

There's nothing I'll be able to do about it, different programs are handling the MP3 differently, and Go PlayAlong may be handling it "incorrectly" by LAME's standards. It doesn't appear most MP3 files will run into this, for some reason, this particular MP3 file is weird. The workaround was to either correct the first beat position in EOF to account for the discrepancy or to use Audacity (which coincidentally seems to treat the MP3 the same as Go PlayAlong) to convert the MP3 to OGG and use that OGG file with EOF (ie. name it guitar.ogg and put it in the project folder).