TO 10 years since its crackling debut, The Elder Scrolls V: Skyrim continues to thrill RPG lovers, especially the more geeky ones.

After embracing the fanciest mods – and the most useful ones – another modder found a way to reuse the in-game voices of Skyrim to create a series of unpublished content, all using machine learning. The result? We can admire it in a fanmade trailer.

SKVA Synth: much more than a simple mod for Skyrim

DanRuta is the modder who gave birth to SKVA Synth , a mod that can be exploited as an authentic speech synthesizer . The tool, based on the use of Artificial Intelligence, is able to generate new lines of dialogue starting from the voices associated with Skyrim NPCs, adopting what the author defines as a “neural speech synthesis”.

SKVA Synth allows users to upload different voice models from various Bethesda open-world games – not just TES V, then. After choosing the desired voices it will be possible to adjust aspects such as the intonation, emphasis and duration of the syllables spoken . To understand the potential of this tool we leave you to a trailer made by the user Adriac.

The voices of some characters of Skyrim (Serana, voiced by Laura Bailey) and even of Fallout 4 (Cait, Katy Townsend) have been adapted to utter unreleased lines , all thanks to machine learning. Of course, there is always room for improvement – the dialogues could be even more natural – but the first result is incredible.

These tools based on the use of AI could be chosen in the future to create any kind of content – or entire video games – using the voices of any voice actor. It will probably take many years before we can replicate the quality of the original products, yet the basics are there and we can’t wait to see new experiments, perhaps with other games.