hl2 overwatch voices

Ron

Developer
Developer
HL2 RP Server Director
Joined
Apr 26, 2016
Messages
413
Nebulae
833
Basic guide of how (I think) valve made the overwatch voices.

Well, a while back in LP I was interested in seeing how exactly Valve made their loudspeaker voice. I (believe I) figured out how they converted it (Besides OTA), but 70% of the voice is the actor.

Now, the batch converter you will need is made by Sony Creative Software, but you can't buy it, you can only download it. As you can guess, you need a key to use it (I'll let you figure out how to get that key :wink:) Download link is http://www.sonycreativesoftware.com/dl/dl.aspx?dwnid=80. Also, to my knowledge there is no other program that can open the batch converter files (Not even other Sony Foundation programs according to their support members).


Outcome: http://vocaro o.com/i/s1wKDyApyUhP Obviously, far from perfect. This was a text to speech robot and the recording cut off something in the middle, but you could tell how that announcement effect was applied.

Great, step 1.
Record the voice. The voice is 70% of the outcome. The actor is Ellen McLain, so anything other than her or someone who can impersonate her won't come out the same way as the other sounds. A big part is also the tone of voice, the pronunciation of certain words, etc. Getting the voice is on you.

1.1 Out of batch editing. This MIGHT be needed, like pitching down the voice and things like that before running it through the converters. Find what works best.

2. Get the Hl2 batch converters. They are in Hl2 Game file--->Sound--->VO--->Batch converters. That or search for .bcs files in the game folder and get them all.

3. There aren't many guides out there for batch converter, but basically you will want to load the script, load the media, change the output settings and then convert the files. It's a really shitty program because of how old it is, but once you get around some of the issues, its alright to work with.

4. Make what you want. Here are the pathways of batch converters I've found to be useful when working with the files.
1. For a loudspeaker voice (Dispatch w/e you want to call it) Convert to 8 bit---> voice to loudspeaker voice.
2. For a radio voice (Dispatch over CP radio) Convert to 8 bit---> preprocess voice normalize--> voice to radio voice.
3. Unit talk. Convert to 8 bit----> voice trim silence start and end. Unit talk is 80% of how the person saying it says it. I haven't really gotten this one to work out well because of that.

Some tips:
 
Reactions: List

Erkor

Narrative/Lore Management
Joined
Jun 15, 2016
Messages
3,011
Nebulae
8,559
unfortunately the cp batch converter sound isnt the one used in the english version, but the one used for diff. localizations

they have a distinctly different radio tone
 
Reactions: List

Oktuberfest

String
Joined
Nov 1, 2017
Messages
1
Nebulae
0
This is over a year old and I apologize for unearthing it, but I figured it wouldn't hurt to ask given the lack of resources anywhere else- and that my research is yielding nothing. @Erkor do you know anything about how to use the .bcs files within sound forge pro 11? Sources state the program can use it, but the scripts within the program are .c, not .bcs- which proves to be a problem.
 
Last edited:

Erkor

Narrative/Lore Management
Joined
Jun 15, 2016
Messages
3,011
Nebulae
8,559
hi im here to correct myself

This entire process (the one to make a good, filtered CP voice, that is) requires a heavy amount of editing and VST usage; I've done it in Cocko's REAPER instead of Audacity because it simply is a better program.
 

welcome to the Retirement

Narrative/Lore Manager
Joined
Apr 26, 2016
Messages
5,171
Nebulae
28,634
i think the reason it's so hard to replicate overwatch vo-coders was because the secret to it's unique sound lays not within editing but in the actor who voiced them. he had a very distinct, overly-bassy yet deep voice. little to no filters (minus the obv radio ones and very minor pitch corrrection) were used iirc.
 

Erkor

Narrative/Lore Management
Joined
Jun 15, 2016
Messages
3,011
Nebulae
8,559
i think the reason it's so hard to replicate overwatch vo-coders was because the secret to it's unique sound lays not within editing but in the actor who voiced them. he had a very distinct, overly-bassy yet deep voice. little to no filters (minus the obv radio ones and very minor pitch corrrection) were used iirc.
in part, but one of the biggest things is the heavily distorted crackle sound that is omnipresent in the voice acting

i've no way to replicate it as every time i try it, it's still accompanied with the actual dry voice sound as well

it's as if they distorted it but removed the input voice completely, leaving only the distortion filter as voice