Dunder

Dunder is the enigmatic head of the Z Music Club. Not much is known about the boy, as it appears all mentions of his identity have been scrubbed from public record. He is often seen wearing an OLED Mask, that typically displays nothing but a question mark.

Dunder very rarely makes on stage appearances, let alone ones in which he sings. Most simply describe his voice as normal or not something they’d commit to memory.

Character Info

Character information pertains to Z Vocal Project lore, feel free to disregard it.

Name: Dunder
Gender: Male
Pronouns: He/Him
Birthday: June 11th
Age: 16
Character Item: OLED Mask
Twitter: @DunderAI
Instagram: @DunderAI
Art Tag: #Dunder_IntercalarisART

Voicebanks

Download on Itch.IO

The download pages for released Dunder voicebanks are over on Itch.io. You can purchase his license on the song voicebank page.
Dunder also has an English and 2 Japanese Coefont banks.

Released Voicebanks

Dunder_CV+ ~Proof of Concept~ (Released April 22nd 2024)
English/Japanese Diffsinger voicebank trained on data from Dunder_CV Act II voicebanks.
Dunder_Aki Act II (Released March 13th 2024)
CV Japanese Soft voicebank meant for use in OpenUTAU. Part of Dunder_CV Act II

Dunder_Genesis Act II (Released March 13th 2024)
Multipitch CV Japanese Dark voicebank meant for use in OpenUTAU. Part of Dunder_CV Act II
Dunder_Natsu Act II (Released March 13th 2024)
Multipitch CV Japanese “Joyful” voicebank meant for use in OpenUTAU. Part of Dunder_CV Act II
Dunder_Haru Act II (Released March 11th 2024)
CV Japanese Power voicebank meant for use in OpenUTAU. Part of Dunder_CV Act II
Dunder α Type C (Released March 7th 2024)
English voicebank for Z Idol Talker with a softer tone. In the middle of recording more data.
Dunder (Released March 3rd 2024)
English voicebank for Z Idol Talker with a normal tone of voice.
Dunder V2 from Z Vocal Project (Released December 27th 2023)
Updated English talk voicebank for CoeFont with maybe an hour of recording data.
Dunder Announcer, Dunder French, Dunder Castilian, Dunder Spanish (Latin) (Released November 9th 2023)
English, French, and Spanish piper models formatted for use with Z Idol Talker. These are test voicebanks that could use more training/data but will only get them if there is demand. Downloadable on the Z Idol Talker itch.io page.
Dunder_Pocket II (Released October 28th 2023)
A kits.ai voicebank trained on newer singing samples of his voice provider than the original Dunder_Pocket. Can do voice conversion and text to speech. Has slight nuances from the original so use whichever you like more.
Dunder_Eon (Released September 9th 2023)
A talk voicebank for UniCoe. Standalone voicebank release, UniCoe will still need to be downloaded from it’s website. Will get a unique design once my drawing tablet is fixed. Core voicebank also released for OpenUTAU as Dunder_Genesis_VCV TalkBeta.
Dunder_Genesis_VCV TalkBeta (Released September 9th 2023)
A monopitch VCV voicebank recorded for his UniCoe, wasn’t intended for singing but it sounds decent enough. Will need to mess with the OTO at a later date.
Dunder from Z Vocal Project (English + Japanese) (Released August 29th 2023)
1. An English Coefont bank, was intentionally given a strange accent, however it needs a lot more data right now lol, the 100 sentence plan is super LQ.
2. A Japanese Coefont Bank. Still suffers from low data at some points but is generally better than the English ones.

3. A Japanese Coefont Bank – Sorrow Append. Honestly sounds about the same as the normal JP bank in terms of quality. Which is to say 3x the amount of data wasn’t enough.
Dunder_Fuyu Act II (Released July 7th 2023)
A multipitch Falsetto CV voicebank meant for use in OpenUTAU. There are currently two pitches, B4 and F4. These pitches are automatically set up for use in OpenUTAU, but if necessary, F4 can be access with the prefix #G.
Dunder_Intercalaris AI BETA (Released April 23rd 2023)
An Enunu Voicebank similar to Intercalaris. Trained on about 1 hour of singing data that was hand labeled. Has some tone labeling that will be improved in future updates, along with timing and clarity.

Dunder_Mercedonius AI ALPHA (Released April 23rd 2023)

An Enunu Voicebank similar to Intercalaris. Trained on about 1 hour and 20 minutes of singing data that was automatically labeled. Has some tone labeling that will be improved in future updates, along with timing and clarity. Most glaringly Mercedonius AI Alpha likes to breathe ALOT because end breaths didn’t get labeled at all.
Dunder_Pocket (Released April 23rd 2023) (Ported to Kits on October 28th 2023)
A Retrieval-Based Voice Conversion voicebank capable of converting pre-rendered singing files into Dunder’s voice! ~~You can also use it to convert your voice to Dunder’s in real-time to mixed results (Please do not use it on Live Streams).~~ It is trained on the same data set as Dunder_Mercedonius AI Alpha, alongside a pre-trained dataset. As of 8/28/2023 the voicebank is only available through Kits.AI, realtime voice conversion is no longer possible but he can now do Text to Speech.
Dunder_Intercalaris α (Released June 11th 2022)
Dunder_Intercalaris α is the alpha version of Dunder_Intercalaris. It is a “Dark” Japanese CVVC voicebank. This version only includes 1 pitches and it’s OTO still needs some fine tuning.

Upcoming Voicebanks

Dunder_Solstice
A light/falsetto bipitch Japanese VCV voicebank. Hopefully releasing soon it is fully recorded with pitch one being G3 and pitch two being A4
Dunder_CV Act II
Updated versions of the CV banks Dunder_Fuyu, Natsu, Haru, Aki, and Genesis. Natsu’s voice type is essentially the same, but it will be called “Joyful” instead of “Normal” now. Releasing soon
Dunder_Intercalaris+
A Retrieval Based Voice Conversion voicebank trained on Intercalaris Samples. Will be released when I feel like it.
Dunder_Intercalaris
Dunder_Intercalaris is a “Dark” Japanese CVVC voicebank. A 2-pitch alpha version has been released. The full version has been put on the back burner to focus on Dunder AI.
Dunder_Mercedonius
Dunder_Mercedonius is a “Dark” English ARPA voicebank. Lowest priority, might be scrapped in favor of only having an AI voicebank for English.
Dunder Sample Pack
A Dunder Sample Pack for use in Music Production. Will be given out to owners of his commercial license exclusively
Dunder_Genesis DIFF & Dunder_Mercedonius DIFF
Japanese and English DiffSinger voicebanks

Retired Voicebanks

Dunder_CV (Dunder_Aki, Dunder_Natsu, Dunder_Haru, Dunder_Fuyu, Dunder_Genesis) (Released in full June 11th 2017)
These were a set a 5 CV voicebanks I recorded when I was like 15, the OTO wasn’t the best and as I was a minor when they were recorded, it was ultimately for the best to discontinue them.
Dunder_Voice (Released June 11th 2017)
This was just a set of me saying phrases for use in song, as with Dunder_CV, I was 15, but also I was REALLY bad at voice acting so the samples were also just bad.

Private Voicebanks

Dunder_Pure

An Eleven Labs voicebank based on Dunder. It cannot be distributed logistically.
Dunder_Talker
A Low Quality TacoTron2 Voicebank based on Dunder. The engine we’re looking at for Dunder’s talk voicebank isn’t TT2 so we stopped development on this.

Demos

Early Alpha Tests

Early Alpha Test of Dunder_Intercalaris AI with about 20-30 minutes of data. Chickened out of making it public

Early Alpha Test of Dunder_Intercalaris AI with about 20-30 minutes of data. This cover will be redone when the voicebank has more data.

Dunder_Pocket singing A Way of Life (ATLUS Kitajoh Remix). Base vocals were the Synthesizer V Voicebank Solaria.

Dunder_Pocket singing Good Morning Sue. Base vocals were separated from the original song.

Dunder_Pocket singing A Lone Prayer. Base vocals were the Synthesizer V Voicebank Mai.

Dunder V2 from Z Vocal Project

Product Commercials

Fun little attempts at making a commercial for Dunder V2 from Z Vocal Project. These weren’t actually aired anywhere.

Dunder_CV Act II

Beta Demos

Beta Demos of Dunder_Genesis Act II and Dunder_Fuyu Act II. OTOs and recordings still need a second pass, Genesis will probably get another pitch.

Alpha Test

Demo usage of Dunder_Intercalaris α, in this demonstration the voicebank is only 1 pitch, and it’s OTO still needs some fine tuning.
The UST was made by Lucifer/Star, and uses their tuning as well.
(This cover will no longer be featured when the commercial license becomes available for purchase)

Publicly released test of Dunder_SVC (Intercalaris+).
The voicebank has around 2 hours of data and 130k steps, however the low range is nonexistent. It still funny to hear him sing tho.

Publicly released test of Dunder_SVC (Intercalaris+).
Tests of the voicebank in Japanese, Chinese, and English. Source data is only in Japanese. Same model as in the Emotions test.
Suki De Itte was rendered over output of an SVP made and tuned by VocaCircus.

Mercedonius 8% Test

Publicly released test of Dunder_Mercedonius.
This voicebank is going to be rerecorded entirely and the OTO shown in this video was only around 8% done. I’m really only keeping this demonstration up as proof that work has actually been done on it.
Regardless, this video should give you a decent idea of what Dunder’s English voicebank will sound like tone wise.