Valheim Voices
English voice bank for Audio Dialogues: 738 spoken lines, eleven voices, from Haldor to the runestones. Synthetic voices.Valheim Voices
The English voice bank for Audio Dialogues: 738 spoken lines, eighty minutes of audio, eleven voices. Install both, this one being the sound and the other one the code.
The cast
| Voice | Lines | Register |
|---|---|---|
| dvergr | 174 | low and gruff, one voice for all four mages |
| runestone | 141 | grave narrator, slow delivery, the voice you hear most |
| fallen | 80 | breathy, a spent timbre |
| hildir | 75 | feminine, warm |
| hugin | 63 | reedy, high, restless |
| dreams | 61 | whispered: dreams and the voices of the dead |
| haldor | 50 | jovial trader |
| munin | 42 | reedy, lower, sardonic |
| shadow | 30 | cold and flat, the shadow people of the Deep North |
| bogwitch | 18 | feminine, aged, quavering |
| saga | 4 | solemn, and the fallback voice |
How the voices were made
Speech synthesis, with Qwen3-TTS. Each voice was written as a description rather than cloned from a person: no real voice was sampled, imitated or reproduced. The description makes one reference recording per character, and every line of that character is then spoken in the timbre of that recording, which is what keeps a trader sounding like himself across fifty lines.
Ogg Vorbis, 24 kHz mono, 25 MB in all. Mono is not a shortcut: Unity only spatializes mono, and a stereo clip would play flat wherever the speaker stands.
What it does not cover
The intro and ending cinematics keep their subtitles, being timestamped against the video. Reading a runestone again from the compendium stays silent, the journal having kept the translated text rather than the key it came from. English only for now.
Replacing a few lines
Voice packs stack. A pack holding a handful of ogg files and a small index.json sits on
top of this one, key by key, without replacing the rest of the bank. The format is described
on the Audio Dialogues page.
