Fallout 2 mod New voice for Cassidy

What do you think of joey's first attempt at a Cassidy voice?

  • Perfect

    Votes: 25 73.5%
  • Decent

    Votes: 9 26.5%
  • It fucking stinks

    Votes: 0 0.0%
  • Needs work, but getting there

    Votes: 0 0.0%

  • Total voters
    34
I don't know a great deal about how animation stuff works in FO2, admittedly, but I would most definitely argue that it's not "the same" situation with new animations + with new audio. With audio, dialogue especially, it's (to me, at least) extremely jarring when the quality of additions is inconsistent with the quality of the originals. It's a very common problem with mods for a lot of games. We will never be able to attain higher quality/clearer sounding voice lines for, e.g., Sulik, or the elder, or Harold or whoever.
Yes, exactly what @Lexx says about animations.
But, never say never in our age of AI/ML advancements.

So I'm of the opinion that anything new added that claims to be part of a "restoration" should try to sit as seamlessly as possible alongside those original performances.
Since it's going to be an independent mod, it doesn't have to claim anything.

Does that make sense?
I get your argument, but I reserve judgement on that until actually trying it in-game.
Anyway, the release will likely have different packages so that everyone can use what they want.

More importantly, how is it that the original recordings that you shared are mono, and those converted are stereo?
 
Last edited:
Just wondering... I'm not that familiar with sound recording, voice acting doesn't benefit from stereo?
 
Just wondering... I'm not that familiar with sound recording, voice acting doesn't benefit from stereo?

Nah, not in general, and not in FO2. Not even in a 3D game engine with its own 3d soundscape really :) music is another story tho
 
Nah, not in general, and not in FO2. Not even in a 3D game engine with its own 3d soundscape really :) music is another story tho

I respectfully disagree.. well sort of.. You see back in the day between 1997 and 2005 there were those fancy Aureal3D soundcards and drivers, which in combination with a game supporting their technology, one with few weeks of training isuch environment could pinpoint the sound source in 3D space (such as footsteps) while playing on stereo headphones. That was achivable thanks to HRTF (head related transfer function). an human brain ability to notice a slight difference in stereo sound channel timing of a frequently repeated sound.
players running on Aureal3D soundcards playing counter strike 1.6 CD/DVD version do notice it while playing on winXP, it was the last windows with DirectSound3D libraries present (up to XPSP2), which enabled this marvel. CS 1.6 on winXP, for Players with Aureal3D cards trained long enaugh did not need fancy player model figurines, like they do have in CS GO nowadays, to figure out where the enemy is, as simply by hearing enemy footsteps they could easily pinpoint a sector in 3D space where exactly the sound is coming from, and that's thanks to custom stereo sound technology and simple stereo headphones. There were other games supporting this technology back in the day, Like older versions of Quake3 Arena for example, nowadays most of them still run on winXP, probably due to owning an Aureal3D soundcard. That is why CS 1.6 players hate Steam version (which cut out win xp support from steam, and A3D soundsystem from the game).
I bet modern time players would love to grab their hands on such technology i modern games, however Creative Labs Technologies, which took over Aureal by suing them, and buying them out, shelved the technology, and when the time came to use it, they couldn't crack it. modern day equivalents of this technology use software implementations such as "Sonic for Windows" however folks do not understand that sound processing requires an independant processor thread and decent calculating power and should be done on a separate sound processor, like Aureal did, just like graphics calculations are done on a GPU not CPU, as efficiency is a lot better on a real thing than virtualization. that's why sonic for windows gives poor performance in games.

anyways.. sorry for the off top, but i just felt it is nessecary to tell this, so that you'd have full picture of stereo sounding in 3D games, as Your statement may be true now, but it wasn't always the case.
 
And now it's endorsed by non other than MCA:



Chris Avellone on Twitter said:
One of the best parts of fan mod projects (in this case, the Fallout 2 Restoration Project) is hearing additions we never had time or budget for - like Cassidy getting his own voice actor, courtesy of Joey Bracken (
@stillliquid
). #Fallout Great stuff!
 
@slick liquid there's a Vic talking head now, look at this: https://nma-fallout.com/threads/rpu-talking-heads-addon.220886/page-4#post-4466674

A lot of new talking heads actually that needs voices.

I just found this thread and wanted to leave a couple comments:

First off I created a seemingly way better tool for doing the lip extraction for fallout. Minimizes the amount of manual adjustments that need to be made. I plan to publish the tool after we complete work on the project in my next comment.

@Goat_Boy and I are putting together voice acting fot all talking heads in his mod right now, including vic.
 
Black Electric do you have a screen grab of the software? Just curious!
@.Pixote. There isn't much to show. The process consists of:

1. Running a python script (Penn Forced Alignment, open source which I found online) to do the phoneme extraction from the wav to a .grd file.

2. (Optional but recommended) Review the extracted .grd phoneme timing in Praat (also free) to ensure the timing looks correct and make changes if needed. (If the audio is clear, usually accuracy is 100%)

3. Use the python script I wrote to generate the .LIP from the .GRD file.

There's really no GUI element to it. It's all scripting at the moment. Maybe if I decide to put more time in it I can make a basic user interface.

There are also a handful of supporting scripts I am writing as well. One is to do the Dialog extracting using the .ssl and the .msg files for a given character, and concatenate lines together that are broken up. One will be to help clip silence at the beginning and end of audio, transform to 22050Hz, and convert to ACM, one will be to update the .ssl file automatically with the updated new dialog line numbers resulting from concatenation (and because some of the lines need to convert the input string to a line number in the .msg to reference the .LIP file), one will be to deploy game files, one will be to update the .msg files...

And there are several tweeks to some of the steps above that I need to make to support batch processing... we are dealing with 1000's of lines of dialog so... hand doing all this is out the window.

As I said, mostly scripting so not much to see, lol.
 
This is a project I had planned on attempting after my current overworld map tool is finished, but it would be awesome if you had something already set up for good lip-syncing!
Can't wait to make a video about this! Please drop something when you're ready :)
 
@QuantumApprentice all I really have to demo for the script is the end product. Below is a Marcus lip sync I put together to test robustness against singing but synthesized voices are harder to perform phoneme extraction on.

Was planning on videoing the process I use when I've completed all the tools so I can demo a start to finish.

 
hey if you want to make the video yourself, then more power to you :P
I was just going to do what I did for the last few videos where I provided a script so people could just run it, which I think is what you're describing.
However you do it, a lot of us can't wait to get our hands on what you're working on! :P
 
hey if you want to make the video yourself, then more power to you :P
I was just going to do what I did for the last few videos where I provided a script so people could just run it, which I think is what you're describing.
However you do it, a lot of us can't wait to get our hands on what you're working on! :P

Yeah, I've edited plenty of YT videos in my time for a separate car-based channel that I run, so I am comfortable with editing tools. Just trying to dial in the screen record settings when playing the game. Videos are definitely the best way imo to show a process like this.

The most complex part has actually been the syntax analysis of the .ssl files to identify the list of lines for the voice actor. Most of them are easy one liners, but some of them use compare statements to determine one of two scripts that a character should say based on the game conditions. It gets more complex when there are multiple branch statements so you go from 1 to 2 to 4 possible voice dialogs in a single GSAY_REPLY statement. Not to mention random statements and then having to swap in a generic pronoun for the player's name that fits the context of who they are talking to.

Then of course there are lines with variable values as part of the dialog itself. I have the script spiking out these lines so that I know to go back and manually tweak the script so that the user says something generic, e.g. instead of saying "I will pay you *x* caps." They might say "I will pay you plenty of caps."

I've attached a 'lineRecord' file as a little show and tell. This is part of what the dialog extractor generates when analyzing the .SSL files. Most of them are easy one liners and maintain the original .msg line number. Some of them though are concatenations of multiple text lines and so I generated new line numbers that will be appended to the .msg file. (Any line number with value 56** is a new line I generated to have the voice actors recite, and will be added to the .msg file) You can see in lines 5602-5604 an example of this branching complexity in the dialog.

Currently in the casting phase but one of the next steps is to make a script to re-write the .ssl file using the new lines. This will be a bit complex when it comes to the branched dialog as I will have to code in branch statements into the file and write the gsay_reply statement with the hard-coded value from the lineRecord.
 

Attachments

Last edited:
Does anyone know where I can find a .DAT explorer that I can actually create the .dat files with? I am using version 1.43 by DIMS but it SUCKS for creating the .dat's. I can't delete folders / files so if I get something wrong, the .dat is screwed and I have to start over.

It looks like there used to be a DAT Explorer 2 at the link below but maybe it's been removed, or I just don't have permission.

https://www.nma-fallout.com/resources/fallout-dat-explorer-ii.121/
 
Back
Top