September 14, 2017 at 4:20 pm #814
So the posts about this issue on the Nexus mod page are threatening to overtake the whole thing, so I figured I’d move the discussion here.
Here’s the original post by Sothaer:
When I have this mod installed, I get very noticeable cutting at the end of dialogue (seems to exist in the vanilla game regardless of what I do but this mod seems to amplify it / create it where it didn’t exist before). Does anybody else have this issue / is there a fix for it?
And there were a lot of replies, but here’s the ones that seem to be currently relevant. First one by me:
Immersion does matter to me, and I hate that that’s happening. I will continue to work on it, but unfortunately I’ll be working blind (deaf?) as I can’t reproduce it. Unfortunately, I think it’s likely that it’ll always happen when I rework the sound file for the various fixes that’s necessary for. Here are the steps I’m taking, maybe someone who has worked with this can provide a clue:
1) Find a .fsb file in the audio directory, such as alistair_main.fsb.
2) Extract all the sound files from it with fsbext. This produces “.wav.mp3” files.
3) Convert the “.wav.mp3” files to “.wav.wav” files using Goldwave Batch Processing.
4) Use a command prompt to rename all the “.wav.wav” files to just “.wav”.
And then use FMOD and so forth to recreate an FSB from those files.
The original .mp3 files when extracted sound fine to me. When converted to a .wav and played in the operating system, they seem to get cut off right before the end (so “circle” in that line gets cut off in the middle) – however, at least for me, when actually put back into the game they sound fine, and you’re all telling me the word does actually mostly complete. Still, it’s probably responsible for what you guys are hearing. Makes me think it’s the conversion from .mp3 to .wav that’s doing it, and I don’t know of any good tool other than Goldwave to convert a mass number of files from .mp3 to .wav. Anyone got another tool for doing that, just to see? Goldwave seems to pretty much be the industry standard for this type of work, though, so it’s troubling…
Illusia later noted:
I was thinking the problem might have something to do with .mp3 compression artifacts. It seem rather silly to extract the files in a compressed format just to convert them back into a lossless format. Or are they actually stored as .mp3s?
Yes, the only two fsb extraction programs I’m aware of (fsbext and FSB Extractor) yield .mp3 files, but the voiceover processing and FMOD require them to be .wav’s. So, so, so incredibly annoying. The resulting .wav files have to be 24000 Hz 16 bit mono. So far, I’ve tried 8 different mp3 to wav conversion tools and none except Goldwave will fully do the job, most of them won’t convert to 24000 Hz (closest is 20500 I think).
One thing I did discover is that only Groove Music (my default player) cuts off “circle” in the middle. If I play the converted .wav files with Windows Media Player in the operating system, they seem fine. But, again, I can’t hear what you guys are hearing at all, so that doesn’t mean anything.
Illusia’s latest reply:
“The FSB file format wasn’t designed to be a format for storing the master copy sound assets. Each time you unpack and repack the data there will be quality loss if the FSB is using any kind of compression.”
…and I suspect that’s at least part of the problem. The original project files get compiled (and compressed) into .fsb –> extracted and then converted in order for you to work with them –> recompiled (and recompressed). And each step results in some loss of the quality. I’m guessing there’s no way to access the original .wav files?
Awawe Audio looks like it should be able to handle the conversion. It’s not free, though.
I won’t be putting money into buying a package to deal with this, I’m afraid. Donations and Amazon portal buys haven’t even managed to pay for this website domain yet. When I ask for support for this mod, something like that *is precisely* the kind of thing I’m talking about… anything that actually compensates me for my time is, like, 3 years away at the current rate… and until there’s enough left over after all the other costs to pay for something like that, I’m not putting anything more than my (apparently destined to be valueless) time into it. I hope that’s understandable.
Boxoft DID seem to improve on what I’m getting from Goldwave (it took almost 10 times as long to convert them, but that’s not a real problem) – at least, it doesn’t make Groove Music playback cut the last word in half – but it insisted on changing the files to a different Hz, rendering them useless.
I did try the trial version of Awawe on the “Circle” line. The resulting file does suffer the same last-word-cut-in-half effect in Groove Music playback. It does have “normalize” options that *maybe* could improve the situation? I don’t know – I am by no means a professional sound editor, and since I can’t even reproduce the problem, I can’t tell if adding that helps in any way.September 14, 2017 at 6:01 pm #816
Sothaer here, ty for making this thread.
So you said you can’t reproduce the problem in your own game (correct me if I misunderstood)? If it’s at all possible and not a waste of time for you, do you mind uploading a clip of you engaging in the Circle Mage vs Alistair dialogue at Ostagar in your own game? W/ music off ofc for clarity. If it truly isn’t there for you, that would be odd and that means there’s maybe a way of fixing it by myself.
Either way I’ll be reading this intently and I hope you’re successful w a potential fix 🙂September 14, 2017 at 6:06 pm #817
That’s a great idea, though taking video of in game stuff isn’t something I have any practice in. Got any tool in particular you’d recommend? I’m not a fan of GeForce Experience but I’ll install it for this if I need to.September 14, 2017 at 6:21 pm #818
OBS is what I used to create the clip I sent you (I streamed it to Twitch.tv, but OBS has a record option that you can use just as well).September 14, 2017 at 7:03 pm #820
Ok cool, managed to get that working. Didn’t feel like signing up for any other hosting, which is fine, since I can do it myself!
It’s an .mp4 download. Black screen for a few seconds, then it starts. Apologies for the jittering at the beginning of the video, that’s me turning 3D Vision off. And… yes. The game loads insanely fast. Built this rig myself. /nerdflex
Let me know if you hear anything odd at the end of the Circle line. I don’t.September 14, 2017 at 10:15 pm #821
Yeah, I do. There’s the same “click” in the audio at the very end of “Circle”. That’s the most audible one, but there are cuts at the end of most of the other sentences for me (varying degrees, some I have to really listen for and aren’t really noticeable under regular circumstances).
I’m unsure if I am just articulating this badly / I have a weird ear for this / your PC genuinely isn’t processing the audio the same way mine is.
Gah I wish I had a better way of describing it haha. Clunky words.September 14, 2017 at 10:21 pm #822
The best way I can think to put it is that it seems the audio file for each set of dialogue is ending like .5 sec earlier than it does in the unmodded game. So that cut in the audio overlaps with the very end of words in some spots and makes it all sound less natural.September 15, 2017 at 10:59 am #825
Tonight when I get home, I’ll post the .mp3 file that is directly extracted from the FSB, and the .wav I convert it to with Goldwave. It’ll be useful to know how audible the issue is at each of those steps in the process.September 15, 2017 at 2:01 pm #826
Cool I’ll keep an eye on this thread, tySeptember 16, 2017 at 12:58 pm #829
Apologies for the delay.
The .mp3 is the mage’s line as extracted from the .FSB. The .wav is after conversion via Goldwave. Curious which step is doing more to create the problem. Thanks for your help!September 16, 2017 at 11:13 pm #830
It’s weird cause both of those sound exactly the same to me, but in-game I can notice a clear difference between the unmodded game and the modded game…
I can hear the cut in those files and in the modded game, but I can’t in the unmodded game
Bleh this is frustrating for me, I can’t imagine how it is for you lolSeptember 17, 2017 at 12:10 am #833
If you’re hearing it in the .mp3, and the .wav is no different, that confirms Goldwave isn’t the problem. That’s how the line is coming out of the .fsb, with the two most well known fsb extractors I found (one was actually developed for Dragon Age and expanded to all .fsbs). So yeah, unless someone can get a hold of the original wavs, not a thing I can do about it. 🙁September 17, 2017 at 12:37 am #834
Actually, there’s something I can do. What I’ll work on is producing versions of Alistair and Morrigan’s dialogues with everything except the Alistair Isn’t King fixes. If you can hear the difference (not everyone can), you can pop these versions into override at least until the Landsmeet, and if Alistair IS going to be king, just leave them thete. If he isn’t going to be king, at least you’ll only have to deal with the sound issues for the last 15% or so of the game. Does that work as a solution?
Oh, do let me know how Sarel and Lanaya’s dialogues turned out when you get to them. They got sound work too.September 17, 2017 at 5:46 am #835
Would it be possible to create new .fsb files just for the dialogues needing audio work?
I found this discussion when searching for info. Sadly, the tutorial is no longer available, but at least it seems to confirm that it should be possible to avoid rebuilding .fsb files and still replace individual audio files.
September 17, 2017 at 9:37 am #837
- This reply was modified 2 months, 3 weeks ago by Illusia.
Funny story. I actually tried something like that initially, and I don’t recall if that version actually made it into the v3.0 beta, but I think it did. The procedure in question is detailed here:
That procedure ends with you only having a secondary .fsb that only contains your new lines, and a new .fev (both created by the FMOD utility which comes with the game) which I believe is the file that contains the relationship, or map if you will, of audio lines in your .fsb’s to the .dlg file. The problem I encountered is that this method SEEMED to work great, and it does 99.9%, but every once in a blue moon, Alistair would deliver the completely wrong line in dialogue, and this could be anywhere in his dialogue file. The mapping in the .fev if you create a “split” .fsb that only contains the new lines doesn’t seem to work 100%. It’s a shame, as it’s otherwise the perfect solution. In the end, to stop the occasional mis-played line, I wound up having to do all of Alistair’s fsb as one unit. I didn’t run into any misplayed lines again after that.
I would work harder on trying to see if I can jiggle the steps to make it work 100%, but the problem is that the misplayed lines are so rare, there’s no way to properly test it. That issue can manifest anywhere in the hundreds of lines in his dialogue. Even a full test run isn’t going to play all his lines, and therefore a mis-mapped audio line could always be missed regardless of how much testing you do.
You must be logged in to reply to this topic.