[WIP] Voices Of Skyrim

[slappy]

Something I'm putting together for myself, but looking for feedback on if it's something worth producing as a "mod" or not. I do a lot of NPC customization for myself, and was struck by the limited amount of voice types in the game. So I fixed that by using DSP on the existing voice sets to make "new" voices using the default dialogue scripts. I've also done custom voice scripting where it didn't exist - for example, children do not have follower dialogue, so I made it up by cutting and pasting their other voice files.

 

Current status:

 

Modified all female voices - created alternate versions of each with a pitch up (younger) and a pitch down (older) version. Modified files have been re-compressed into fuz format to save space.

Built a child follower script - most dialogue has multiple random responses so you don't always get the same answer (after I did this, was seriously tempted to to do that same thing with all default follower dialogue).

Setup as a esm master resource file, so it's easy to plug the voices into a mod.

Tested in game - all new voices are available and working.

My personal version also includes selectable Japanese and French voices, but would probably pull them out of a mod release if I decide to do one (although having a group of bandits yelling at you in a "foreign" language adds a really unique element that the game is seriously lacking in my mind).

 

To do:

 

Add new male voices

Make scripts for other voice types that don't have follower capabilities (yeah I know, there are a few earlier attempts at this part, but they appear to have been abandoned unfinished).

 

Question is what to do with this? It really ends up being a mod resource like the hair packs - unless it gets packaged with an NPC overhaul mod or something similar, it's not much use to 99% of the Skyrim community. And also wondering what type of conflicts I would see with other mods.

 

[lmstearn]

I think it's a great idea- if nothing but a resource. What software have you been using, and how does this cope with accents, phonemes, intonations and emotions. If we can make Text to Speech work, then it's well on the way to a Creation Kit for Sound!

[slappy]

Unfuzer for the extracting and repacking of wav and lip files. Been using Audacity for the repitching of the wav files once extracted. It has a utility to batch process, although it's really buggy with large numbers of files. So I have to do the processing in batches of around 4-500 at a time, so it takes awhile.  I don't alter any of the other related files - even the lip files can be re-used, because the pitch change doesn't affect the length of the file to any significant degree (milliseconds). Pitch changes greater than around 2 semitones tend to have audible artifacts in Audacity, not sure if that is true in other programs.  But it's still great fun to play around with bigger changes, you can make some really strange sounding characters.