PG Music Home
And again it has happened, I generate a song in BIAB, export the tracks to Sonar, get about half way through the song and find BIAB has generated a major chord when a minor chord is noted...and then to compound it generates a minor chord when a major chord is noted.

These are simple triads people, nothing flash...why does BIAB continually do this. It REALLY adds wasted and unneccesary time to create songs. Talk about an inspiration killer mad

It has happened many times before, but enough is enough. Why cannot this program get chord generation correct? ESPECIALLY when it is simple major or minor triads, AND the tracks are set to simple arrangement only.

Really annoying!
Your email doesn't mention any details, such as what chords, what Realtracks etc. so we couldn't fix it based on that.

If you would like the issue to be investigated/fixed, please freeze the track, and email the song file to support@pgmusic.com requesting that they forward it to development and reference this thread.

If you look at the thousands of messages on this forum, you will see a few people with similar issues, but not more than that. This is because we aim to investigate/fix every one that gets reported.
Thanks Peter - I can do that.
But it happens on a frequent basis, so does PG want to continually get these files?

And the fact that it is a random generation thing does not help. IE I can re-generate a song and it generates correctly, then upon another re-generation it does the minor for major (and vice versa) again. Some song parts I have had to re-generate 4 times and then freeze. In the end it gets it right, but it really should not take that much effort imho.

I was thinking of using the app live, but if this is something that can not be fixed I don't think it would be possible.

PS: I didn't mention chords as it is simply a major chord being generated when a minor chord is noted...I would not have thought the chord pitch would be vital?
If past experience holds, they would also need to know the specific RealTrack(s). The frozen song will tell them that.

I have only seen this issue reported maybe four times in the 8 years or so of having RealTracks, and they fixed it right away every time with a patch. It has NEVER happened to me, but I probably use different genres than you so I haven't hit that specific RealTrack. Give them what they asked for so they can research and fix the problem.
PE,

The below is just my take how these problems might come about. I have no particular understanding of the Realtrack process other than playing around with User tracks.

To understand how wrong chords can occur....

It is necessary to understand how BIAB and Realtracks work to appreciate how easy it is for a chord to be: (a) generated correctly sometimes and not other times; (b) how a minor/major mix up might occur.

When Realtracks are created, the musician making them creates a single file where s/he plays a few bars of one chord in different style variations. The musician then moves on to the next chord in the sequence for a few more bars. The end result is that the Realtrack audio file is one continuous file that contains many chords in many keys. Inasmuch as I understand the process, the location of each chord in the audio file is referenced in a separate file.

When a backing is assembled, BIAB references the co-ordinates of the chord entered on the chord sheet and then copies the appropriate section from the original audio file. (Chords are not built up on a note-by-note basis as would be the case if a midi backing was used.) If a chord's reference co-ordinates are slightly inaccurate, it's possible that a wrong chord could be copied. By freezing tracks and sending a note to PG Music about the problem and which bars to listen to so that they can hear it, you help track down the inconsistency so that it can be remedied.

The reason that backings can sometimes be right and sometimes wrong is because of the different variations that the musician played. For example, there might be 10 variations of a C major chord. Of these 10 variations, 9.5 may be referenced correctly. Because BIAB randomly chooses a variation when creating a backing, the program will not always select the inaccurately referenced region. Again this is why freezing and sending the file to PG Music Support (along with problem details) is very useful.

As Peter said, this problem is not a common one but it does sometimes happen. Over the years, I have tested heaps of Realtracks and have only sent four or five frozen SGU files to PG Music with notes about a chord issue. These issues have always been resolved.

Hope the above helps with a little more insight.

Regards,
Noel


))) And the fact that it is a random generation thing does not help

It doesn't matter, because you freeze the file, which forces the same generation, and we hear what you hear.

You haven't told us anything about this Realtracks, so I haven't a clue as to what the issue is. Remember, there are over 2,500 Realtracks.
Okay I'm on a break now so I can add a bit more...sorry for the delay

The style is _GRESROB.STY And the RT is 372 Guitar Acoustic Strumming Hank Sw 120.

I just did re-gen and it did it again. Chord entered is Ebm and it throws out a Eb major. I have frozen the track...now how do I send it to you? Is it from within BIAB, or do I need to export and send via email?

Thanks

PS: @ Noel - cool, thx that was a good explanation!
You would save and the. email then BIAB song which is an sgu file. I was unable to reproduce your issue on a song that had lots of Ebm and used that #372 RealTracks.
Check the file date of c:\bb\rAcG02.st2 it should be Dec. 17, 2013 - if its older, then that would be the issue. If it is the same, then would need to wait until that song arrives.
Thanks Peter, file date matches.

I'll send the file to the address you wrote above, along with a reference to this thread.
Done!
Interesting. I don't think I've ever had this occur, and if I did, I don't recall it.

If you have Melodyne Editor, the fix is quick and simple. ME will let you edit individual notes in a chord of multiple notes, so moving the major 3rd to a minor 3rd and saving it takes a few seconds.

However, if this is truly a bug or glitch, PG needs to find it and fix it. It could however, be unique to your computer based on the specs and components and drivers in your system.

It'll be interesting to see how this shakes out.
As I mentioned, I never had it happen with RealTracks either. But if it is one of these infrequent programming problems, they will fix it in a patch. It's never been a big deal.
If anyone is interested I will post the result of tech supports conclusions? TBH I think it has nothing to do with PC specifications....one thing I noticed was the "glitch" happens where there is a part marker. Not sure if that can throw off the chord generator or not, but I am sure the tech guys will notice it too.
Yes, please post. This will help us give suggestions next time.

Part markers? Very interesting. Really good clue!
Definitely provide the feedback here, it helps countless others (including me)
Well, the file is now with PG Tech and development they confirmed it is there. Will let you know when further info is to hand.
I am a bit late noticing this thread. I guess that is because I am using BIAB more because I am organizing jams with this product and it is becoming more important that BIAB gets it correct.

Regarding this problem in BIAB 2012 I have heard real track chords that do not sound good at all so it is kind of nice to see this thread and know it was not just me. In fact they sound so bad that the midi is far better. Not only that the program generated it consistently (not random). In the end I just switched it back to midi but that was because I was just using the program to practice and I did not care if I heard real instruments. Now that I am aware that it might just be the program and not me maybe I can help if it happens to me again. On the positive side, I have not heard anything yet in BIAB 2016 I don't like. Maybe these problems are solved. However I have been busy writing a web based software to organize the jams and it is finished now. Today is the day I start to use BIAB more often. In the mean time I have some programmer type of thoughts.

Being a programmer if I was knowledgeable of the BIAB program I would create a feature which is a simple button click to save the generated file as well as the original file to some work area and automatically ship it off for inspection by the tech people. This is the kind of effort that solves these types of problems more efficiently. However it might be that this should be restricted to somewhere on the page that shows the actual notes of the chord so people who do not have the ability to hear and know exactly what the wrong chord is can tell it is the wrong chord by the notes being shown on the staff. So (assuming these notes on the staff are actually the notes being played) the program should be able to find this problem itself (a self debugging feature and programming department managers love this). The way I see it if the programmers are able to create a program like this they certainly are able to create such a self debugging feature (maybe a feature that is run before the program gets shipped out). It may not be that easy to do this but I think it is an investment that would pay off for them in the long run. This idea is basically a variant on the verification feature you see in many file backup programs. It does not find the actual bug in the code but it finds the problem it created and certainly helps create more effort to find the solution. However if the notes being shown match the chords the person entered and only the sound is wrong then this makes it more difficult (I am thinking the tech people need to be able to hear that the chord is the wrong chord). So in this case for anyone sending the details of this problem out to the tech people it would make sense to take some screen shots of these notes indicating the exact bars where the problem is. Actually, even if the notes on the page do match the chords that the user entered maybe they can still write a program to compared the input and generated output files automatically. I could be wrong but I understand there is very expensive software that can figure out the exact notes of a chord being played.

Yes John, I grudgingly have to admit that sometimes midi is better! Lol
PG has never resolved the issue that Held chords (i.e. x...) using RT Synth tracks play wrong notes all too often so I've gone to ignoring them now and finding a similar midi style that has, say, M49 or M50 Strings in its Style and using that track instead. Shouldn't have to!!!
Ian
We looked into that issue, and found an instance where an Ebm chord would get played as a Ebmajor. This just affects that RealTracks #372
The fix will appear in the next patch, but if you want to get the fix now, download these tw files to your c:\BB folder

ST2
https://www.dropbox.com/s/wpwqrzkmws5f6jk/RAcG02.ST2?dl=0
XT2
https://www.dropbox.com/s/0o83a4727fk21g7/RAcG02.XT2?dl=0


There are 5 million bars in the 2,500 hours of RealTracks, so it's not surprising that we found 2 bars that had wrong chords. It's easy to fix them if someone spots an issue. Thanks for reporting it
Quote:
There are 5 million bars in the 2,500 RealTracks


Wow! That is an incredible number. It really puts the 4 or 5 instances of mis-mapped chords that I've found into perspective!

Thank you for the quick fix, Peter smile
John, you need to define "wrong chord". If it should be major and it's playing minor, obviously that's a wrong chord. But, if you wanted a maj9th and it's plays a maj7th or a maj6th that's not a wrong chord. That's the virtual band deciding by using it's internal AI to play the maj6th and not the 9th. Or, to add an 11th or whatever if it's a jazz style. There are controls to help with that. Simplify the bass, pedal bass, jazz down or up, etc. Those are not wrong chords, those are you hiring the wrong players for your session! Try some different RT's. Just like real life, hence the name Band in a Box. Feed the guitarist more pizza and beer and you'll get different changes...

The problem with Real Tracks involving chording instruments is the RT file sizes are limited. The players will record all the basic major and minor chords with some extensions in three keys then the program can handle transpositions. Great, but which inversions? There is no way a piano RT can have 4 different two handed inversions of all those chords. Maybe the recorded maj7th just sucks in your key so it uses a 6th or dominant 7th. Look at all the different ways to play chords on a guitar too. Sometimes just changing the key can make the difference.

To record all those combinations would make the RT huge so they have to make compromises and use digital tricks to make it sound good. Most of the time it works well but sometimes it just doesn't. Just like hiring a real band to record your stuff for you. You have an idea in your head but the players just can't quite get it right. Sometimes it's you being too demanding and sometimes it's them.

Midi doesn't have those limitations. You can edit a midi file to play EXACTLY what notes in any instrument track. But midi loses a lot of that live ambience and so it goes, round and round and...

Bob
Originally Posted By: PeterGannon
We looked into that issue, and found an instance several instances where an Ebm chord would get played as a Ebmajor. (fixed haha!)

This just affects that RealTracks #372
The fix will appear in the next patch, but if you want to get the fix now, download these tw files to your c:\BB folder

ST2
https://www.dropbox.com/s/wpwqrzkmws5f6jk/RAcG02.ST2?dl=0
XT2
https://www.dropbox.com/s/0o83a4727fk21g7/RAcG02.XT2?dl=0


There are 5 million bars in the 2,500 hours of RealTracks, so it's not surprising that we found 2 bars that had wrong chords. It's easy to fix them if someone spots an issue. Thanks for reporting it


Thanks Peter. Just for the record I have had it happen several times in the past, but I have not bothered to report it before. This time was the camel haha!

From now on when it happens I will follow the same process and send it in!

Cheers
PS: And thx for the links smile
Originally Posted By: Noel96
Quote:
There are 5 million bars in the 2,500 RealTracks


Wow! That is an incredible number. It really puts the 4 or 5 instances of mis-mapped chords that I've found into perspective!

Thank you for the quick fix, Peter smile

+1
Pax,
And the nice thing is that it gets fixed for everyone when you send it in and we fix it.

BowlesJ
Good idea. About the in house chord tool. We have a number of in house tools, so hopefully can add that one too.
Originally Posted By: PeterGannon
We have a number of in house tools...


Yes, I've noticed that over the years but the staff is much better now.

Sorry, I just couldn't help myself this was too perfect.

Bob
Originally Posted By: PeterGannon
Pax,
And the nice thing is that it gets fixed for everyone when you send it in and we fix it.


gotcha! cool
© PG Music Forums