Skyrim modders are getting creative with this AI text to speech tool

The magic of AI-powered text to speech continues as The Elder Scrolls V: Skyrim modders get their hands on an application that generates speech from text based on voice samples from Bethesda games. It's already inspired a trailer with AI-only voice acting and a new mod that gives Skyrim characters lines to compliment your dragonborn's naked figure. Oh, the things modders will do.

The voice lines are created with an application called SKVA Synth by Dan Ruta which outputs speech based on voice samples from Bethesda games. Ruta published a couple different versions of the application for different Bethesda RPGs. Some of the lines created for this fan made Skyrim trailer sound a bit artificial in pacing and pronunciation, but a few are quite convincing.

Ruta explains the technical side of SKVA Synth on the application's Nexus Mods page, saying that it "wraps around FastPitch [1] models trained on datasets compiled from in-game voice acted lines. The strengths of this model are in the artistic control over the generated audio. Once you generate the audio from your text prompt, you can adjust the pitch and durations using the editor."

As Ruta points out, there are quite a few possible uses for the application. It could be used to create voice lines for Skyrim companion mods. It could also be used to create additional lines for existing quests or characters. It's already been used for this mod that has Skyrim characters compliment you if you're walking around in the nude. They also sound just a bit off to me, but again, extremely neat!

As for future plans for the tool, Ruta says "generally the plan is to keep going down the fairly long list of voices remaining, in Bethesda's games. I do plan on returning to some the voices already released to improve them with further/re-training."

Between this and that other AI speech tool 15.ai—which is also planning to add voices from Bethesda games—I feel like we're in for an interesting year in Skyrim modding.