This mod replaces any vocal instances of Zapper with Taser.
Most lines were re-recorded to match the original, the remainder were uniquely made for this mod to reduce voice line repetition.
Preview Videos
Credits
Individuals
Spoiler!
Authors
Higgue
Dataset Extraction, Mod Compiler, Additional Vocal Coaching
Allen Scott
Input Vocal Performances, Audio FX (Rebecca's Megaphone FX)
DarkShift
Audio FX Assistance (Shade's Radio FX)
AI Training Dataset Source
PAYDAY 3
Intellectual Property
Starbreeze Studios
Developing Studio
Overkill Software
Developing Studio
Deep Silver
Publisher
Character Voice Actors
Simon Kerr
Dallas
Damion Poitier
Chains
Pete Gold
Hoxton
Nicklas Berglund
Wolf
Derek Ray
Houston
Siu-see Hung
Joy
Rebecca LaChance
Pearl
Aoife Duffin
Clover
Kosha Engler
Shade
Jill Winternitz
Rebecca (FBI Van)
Derek Hagen
Cop 1
Henry Garett
Cop 2
Mike Bodie
Cop 3
Jason Keller
Cop 4
Software
Mangio-RVC-Fork
AI Training
MMVCServerSIO
Realtime AI Voice Recording
Voice Model Details
Spoiler!
Dallas
Model inference was done in realtime, feature index was not utilized.
1950 Epochs worth of training.
Dataset was 7 minutes and 46 seconds of speech. (All Assault VO)
Chains
Model inference was done in realtime, feature index was not utilized.
800 Epochs worth of training.
Dataset was 4 minutes and 32 seconds of speech. (All Assault VO)
Hoxton
Model inference was done in realtime, feature index was not utilized.
850 Epochs worth of training.
Dataset was 5 minutes and 27 seconds of speech. (All Assault VO)
Wolf
Model inference was done in realtime, feature index was not utilized.
2600 Epochs worth of training.
Dataset was 9 minutes and 19 seconds of speech. (All Assault VO)
Houston
Model inference was done in realtime mostly, one or two were done through Mangio-RVC-Fork. The feature index was utilized for those few.
1560 Epochs worth of training.
Dataset was 9 minutes and 54 seconds of speech. (All Assault VO)
Joy
Model inference was done in realtime, feature index was not utilized.
1125 Epochs worth of training.
Dataset was 4 minutes and 45 seconds of speech. (All Assault VO)
Pearl
Model inference was done in realtime partially, the rest were processed in post using Mangio-RVC-Fork. The feature index was utilized for most of the lines.
1600 Epochs worth of training.
Dataset was 4 minutes and 2 seconds of speech. (All Assault VO)
Clover
Model inference was done in post using Mangio-RVC-Fork. Feature index was utilized at a ratio of 1.
1800 Epochs worth of training.
Dataset was 10 minutes and 6 seconds of speech. (All Assault VO)
Shade
Model inference was done in realtime, feature index was not utilized.
1625 Epochs worth of training.
Dataset was 7 minutes and 45 seconds of speech.
Radio filter was added in post by DarkShift using Ableton.
Rebecca (FBI Van)
Model inference was done in realtime, feature index was not utilized.
625 Epochs worth of training.
Dataset was 5 minutes and 20 seconds of speech.
Prior to training, Mangio-RVC-Fork was used to DeReverb the speech.
Reverb and Echo reapplied in post by Allen Scott using Audacity.
Cop 1
Model inference was done in realtime, feature index was not utilized.
500 Epochs worth of training.
Dataset was 7 minutes and 41 seconds of speech.
Cop 2
Model inference was done in realtime, feature index was not utilized.
1225 Epochs worth of training.
Dataset was 7 minutes and 52 seconds of speech.
Cop 3
Model inference was done in realtime, feature index was not utilized.
1050 Epochs worth of training.
Dataset was 7 minutes and 19 seconds of speech.
Cop 4
Model inference was done in realtime, feature index was not utilized.
625 Epochs worth of training.
Dataset was 11 minutes 11 seconds of speech.
👊😎