tigermeat
【VOSIM: Release】Hello, Im VOSIM【Chipspeech Original】
updated
USTx/Cover/Mix: Me bitch
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
All TGM DiffSingers have been updated: https://tigermeat.xyz/#tiger-ds
(TIGER, TRITON, Canary & DANROU)
Download USTx: https://tigermeat.xyz/#seqf
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
hey it's me, ladygagacovermaker-p
i mean it's like if i didn't cover this song i'd be a fool.
LOTS of awesome DiffSinger developments happening, this update adds support for German as well as a lot of new training parameters that make the voices a lot more realistic.
I had the idea for this video and it made me laugh really hard so this is what you're getting!!!!!!!!
Vocals: DANROU DS v108
Cover Production/USTx: tigermeat (me) / https://bsky.app/profile/tigermeat.xyz
DANROU DS Development: tigermeat (me)
Illustration: Guillotama / https://bsky.app/profile/guillotama.bsky.social
~*~*~*~*~*~*~*~*~*~*~
DANROU for DS is available to download now! https://tigermeat.xyz/#danrou-ds
USTx download: https://tigermeat.xyz/#seqf
~*~*~*~*~*~*~*~*~*~*~
DANROU for DS is released!!!! This song was really fun (but kinda challenging ) to cover!!! It mainly uses his Light vocal mode, which is more airy and younger sounding, which worked well for a song that was for Kaai Yuki originally!
I hope everyone enjoys him, he was so fun to revive and I'm so glad he's back!!!
Vocals: DANROU DS v108
Cover Production/USTx: tigermeat (me) / https://bsky.app/profile/tigermeat.xyz
DANROU DS Development: tigermeat (me)
Illustration: Guillotama / https://bsky.app/profile/guillotama.bsky.social
~*~*~*~*~*~*~*~*~*~*~
DANROU for DS will be available soon! Read more about him here: https://tigermeat.xyz/#danrou-ds
USTx download: https://tigermeat.xyz/#seqf (I made this one like in 2014 and i'm just re-releasing it)
~*~*~*~*~*~*~*~*~*~*~
This demo shows off DANROU's "Standard" voice mode, and his English support! He has a bit of a lisp, and a cutesy pronunciation. Ponyo is one of my favorite movies of all time, and this song is very silly and cute and perfect for a voice like DANROU.
DANROU will be able to support 8 languages on release: English, Japanese, Chinese (Mandarin), Korean, Thai, Spanish, French, and Russian! You can activate those languages by using the corresponding Phonemizer in OpenUTAU.
I have one more song to release with DANROU, he should be releasing in the next week or so!
Vocals: DANROU DS v108
Cover Production/USTx: tigermeat (me) / https://bsky.app/profile/tigermeat.xyz
DANROU DS Development: tigermeat (me)
Illustration: Guillotama / https://bsky.app/profile/guillotama.bsky.social
~*~*~*~*~*~*~*~*~*~*~
DANROU for DS will be available soon! Read more about him here: https://tigermeat.xyz/#danrou-ds
USTx download: https://tigermeat.xyz/#seqf (I made this one like in 2014 and i'm just re-releasing it)
~*~*~*~*~*~*~*~*~*~*~
DANROU YEAHHHH!!!!!!!!! This was a song I covered with Danrou's original VB wayyyy back in like 2014. I'm so happy with how his design and artwork turned out, done by my good friend Guillotama :3
I'm gonna try to release Danrou before 2025, but I've been having a lot of health problems, so my energy after my dayjob is veeeeery little. This cover took days and days and days just cuz i had to do it in really small increments, but i'm still trying!!!! When Danrou releases, I'll also be updating TIGER, TRITON and Canary. Thanks for watching :3
TIGER/Cover/Video/USTx: tigermeat
-------------
Download TIGER for DiffSinger: https://tigermeat.xyz/#tiger-ds
Download USTx (Unedited): https://tigermeat.xyz/#seqf
**NOTE** USTx is made with English input and phonemes to give it that authentic feel. Won't be automatically compatible with Japanese Voicebanks unless they support an english phonemizer!
-------------
hi i heard this song and thought tiger would slay i hope you enjoy this cover and ustx
apparently this song came out on a gumi album 3 years ago isn't that crazy
Original Song: Lady Gaga / youtube.com/watch?v=Sih0v01Wbek
TIGER/Cover/Video/USTx: tigermeat
Video Assets from Pixabay
-------------
Download TIGER for DiffSinger: https://tigermeat.xyz/#tiger-ds
Download USTx/SVP (Unedited): https://tigermeat.xyz/#seqf
-------------
so. this has been on loop since it came out. litewally.
idk i wanted to make a fun cover of this song and TIGER is ALMOST perfect for it. The highest notes are a bit TOO high for him, but like it's okay because i think i made it sound good anyway. This has to be some of my best tuning ever. I literally used all of TIGER's voice modes in this cover lmao.
THE VIDEO WAS ALSO FUN!!!! - until i had to remake it cuz i didn't vet the original asset i used and it ended up being AI generated.... the video is still decent but it just doesn't feel the same and i'm sad taht I didn't properly vet the asset i used. I learned a good lesson. FUCK Generative AI so fucking hard god i hate it so much.
UST: まきゅさん / youtube.com/watch?v=P8bbl8_-7w8
DANROU for DiffSinger/Cover/Illust: tigermeat
-------------
Read more about DANROU, and listen to raw samples: https://tigermeat.xyz/#danrou-ds
-------------
HES BACK BITCHES!!!!!!!
i've wanted to revive Danrou for SO Long, he holds such a big place in my heart, and I am so happy I finally took the plunge and got him on DiffSinger!
I tried for a really long time to be able to do his voice again, close enough to get actual AI data, but unfortunately i recorded him while my voice was much higher (in 2013 and 2015 respectively). So, his DiffSinger uses 3 of his old UTAU voicebanks.
DANROU for DiffSinger Vocal Modes:
Standard - Based on his Standard VCV voicebank. Solid, twangy tone, predominantly featured in this song!
Soft - Based on his Soft VCV. Very soft and surprisingly consistent! This VM is at at least 50% in this entire song, turned up here and there.
Light - Based on an unreleased 2 pitch CV voicebank, it's much younger sounding and definitely the cutest voicebank.
DANROU for DiffSinger will be releasing pretty soon, with art done by my bestie Guillotama from TasteLoid!!!
~*~*~*~*~*~*~*~*~*~*~
Original Song: Rachel Zegler / youtu.be/aaEFiE7GcAs?si=I9NormlqSz0uWfZ3
Vocals: TRITON DS
Cover Production/USTx: tigermeat (me) / https://x.com/tigermeat_
TRITON Development: tigermeat (me)
Illustration: JulieRaptor / twitter.com/ChulieChu / @chulie_chu
~*~*~*~*~*~*~*~*~*~*~
TRITON is now available for download here! https://tigermeat.xyz/#triton-ds
USTx download: https://tigermeat.xyz/#seqf
~*~*~*~*~*~*~*~*~*~*~
HE'S OUT!!!! I'm so glad this day has finally come!!! TRITON has been such a fun voice to develop and make demo's for and I'm so glad I could release him finally!
This song was a request from his voice provider/my fiance, Ryan. At first I was kinda like "this song???" but after making the USTx I was like.... oh he fits this amazingly. This was one of the first songs I ever made TRITON sing, so I wanted to make it the demo he releases with!
Vocals: TRITON DS (Release date 6/30 TOMORROW!!!)
Cover Production/USTx: FerretFather / https://x.com/ferretfathervs / @JaimeSynth
TRITON Development: tigermeat (me)
Illustration: JulieRaptor / twitter.com/ChulieChu / @chulie_chu
~*~*~*~*~*~*~*~*~*~*~
TRITON will be available on 6/30/24! TOMORROW!!! https://tigermeat.xyz/#triton-ds
USTx download: https://tigermeat.xyz/#seqf
~*~*~*~*~*~*~*~*~*~*~
DEMO 4 OF 5!!!! This one was made by FerretFather, thank you so much!!! It shows off his low range a bit more than other demo's I've shown. TRITON comes out TOMORROW!!!! GET READY!!!!!
Vocals: TRITON DS (Release date 6/30)
Cover Production/USTx: tigermeat (me) / twitter.com/tigermeat_
TRITON Development: tigermeat (me)
Illustration: JulieRaptor / twitter.com/ChulieChu / @chulie_chu
French Pronunciation Check: Hibya / twitter.com/Tweek_Twix / @TweekTwix
~*~*~*~*~*~*~*~*~*~*~
TRITON will be available on 6/30/24! https://tigermeat.xyz/#triton-ds
USTx download: https://tigermeat.xyz/#seqf
~*~*~*~*~*~*~*~*~*~*~
DEMO 3 OF 5 BAYBEE!!!!! This one shows off TRITON's french capabilities, but he's still able to sing in like 40000 other languages with XLS :3
Thanks for checking the pronunciation for me Hibya 💖
Vocals: TRITON DS (Release date 6/30)
Cover Production/USTx: Guillotama of @Tasteloid
TRITON Development: tigermeat (me) / twitter.com/tigermeat_
Illustration: JulieRaptor / twitter.com/ChulieChu
~*~*~*~*~*~*~*~*~*~*~
TRITON will be available on 6/30/24! https://tigermeat.xyz/#triton-ds
USTx download: https://tigermeat.xyz/#seqf
~*~*~*~*~*~*~*~*~*~*~
Here's demo 2 of 5 for TRITON's release, made by my good friend Tama! It turned out really pretty, and we watched this movie recently and it's really weird????? But i liked it. Please enjoy the cover :)
Vocals: TRITON DS (Release date 6/30)
Cover/Mix/USTx/TRITON Development: tigermeat (me) / twitter.com/tigermeat_
Illustration: JulieRaptor / twitter.com/ChulieChu
~*~*~*~*~*~*~*~*~*~*~
TRITON will be available on 6/30/24! https://tigermeat.xyz/#triton-ds
USTx Download: https://tigermeat.xyz/#seqf
~*~*~*~*~*~*~*~*~*~*~
HE'S FINALLY READY EVERYONE!!!!!
TRITON has been in the works for a while, and he's finally ready to go! He's an English-focused DiffSinger with the ability to sing in a couple other languages, and here we're showcasing Japanese! I've always wanted to cover this song, and TRITON's voice is the perfect match!
I would describe his voice as "baritenor", he's kind of in the middle. He focuses a lot more on indie-style music, with a very loose pronunciation. This demo utilizes his "Deluge" voice mode, which is a weak & soft type of voice! You can see some more technical information about him below!
TRITON for DS (v106)
Length of Data: 54 minutes
Recorded Language: English
Supported XLS Languages: Japanese, French, Spanish, Italian, Mandarin Chinese, Korean
Voice Provider: Ryan M. (ilusm bby)
Voice Modes: 4
Gale: Standard singing voice
Deluge: Soft & Calm singing voice
Tempest: Solid & Powerful singing voice
Flurry: Light & Airy singing voice. Focused on the Falsetto range.
SVP: @K3R_A__ / youtube.com/watch?v=Xv6uEyd4dXI
Cover/Tuning/Mix: Me / twitter.com/tigermeat_
MMD Model: AlexFloareaVT / @AlexFloareaVT / twitter.com/AlexFloareaVT
Video Credits:
MMD Motion
Choreography : REIMI KUSAMA (RSC MASAMI STUDIO)
Motion Capture Studio : 株式会社 LOGIC&MAGIC
MMD Motion : Suzuki Rin
Copyright : 1st PLACE Co., Ltd.
MMD Camera by ひな
Cyber Square stage by @mitsuran5
Plug-In Shader by Joshua
MotionBlur3, AutoLuminous4 by Soboro
ikBokeh, ikClut (edits by くろくまの人), ikWetFloor and Chromatic by Ikeno
HG_Shadow_004 by HariganeP
KiraKira v2 by BeamManP
PostHemming by わたり
LightBloom by Rui_CG
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
.: DOWNLOADS!!! :.
Download TIGER's MMD Model & TIGER DS v106: https://tigermeat.xyz/#tiger-ds
Download OpenUTAU Lunai Edition (Recomended for using TIGER!): github.com/keirokeer/OpenUtau-DiffSinger-Lunai/releases/tag/0.1.0.1
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Y'all.... have NO idea how excited I am about this!!!!! Alex Floarea decided to make a TIGER model out of the kindness of their heart.... and it's SO unbelievably good. I've truly never seen a model that has this big boy body-type without looking kinda... wonky? They blew it out of the park and I'm so grateful and happy... please go subscribe to them and follow them right now or else I will show up at the foot of your bed and chant something spooky for 20 minutes and then eat all of the food in your fridge... I'm not kidding
ANYWAY this is also a great time to show off TIGER's new DiffSinger update! This is version v106, yeah I skipped like 2 but I have good reasons. Mostly cuz I did train v104 and v105 but I never released them cuz they weren't worth putting out a whole new release for... BUT this model uses the new reflow backend and has a really stable, amazing pitch model!! Canary has also been updated....
USTx/Tuning/Mix/Video: Me twitter.com/tigermeat_
TRITON voiced by my fiance Ryan M.
TRITON's illustration is being done by JulieRaptor! twitter.com/ChulieChu
~*~*~*~*~*~*~
TRITON will be released soon! Keep an eye out for a release date 👀
My website: https://tigermeat.xyz/
My Sequence Files (SVP/UST/VSQx): https://tigermeat.xyz/#seqf
~*~*~*~*~*~*~
Hi friends! Here's a sample of TRITON. He fucking owns this song, and he's sooo close to release!
~*~*~*~*~*~*~*~*~*~*~
Original Song: Siinamota (Rest in peace king 💖)
SVP: Mina Moonrise ( youtube.com/@minamoonrise ) & Me Tuning/Mix: Me ( twitter.com/tigermeat_ )
~*~*~*~*~*~*~*~*~*~*~
Download TIGER: https://tigermeat.xyz/#tiger-ds
TRITON is not out yet, but he will be soon (hopefully)
Download SVP: https://tigermeat.xyz/#seqf
(download includes VSQx & USTx)
~*~*~*~*~*~*~*~*~*~*~
AY
Mina and I love this song and decided to make a cover. TIGER and TRITON take turns singing, and Canary is background vocals.
Cover/USTx/Mix: Me \ twitter.com/tigermeat_
~*~*~*~*~*~*~*~*~*~*~
Download TIGER: https://tigermeat.xyz/#tiger-ds
Download USTx: https://tigermeat.xyz/#seqf
~*~*~*~*~*~*~*~*~*~*~
Dancing 💃👯🕺 Walking 🚶👟😳 Rearranging Furniture 🛋️🪑🛏️
Literally didn't know this song existed until a few months ago but damn i LOVE IT!!! Enjoy my fast, very shoddy video editing skills lol.
USTx: Came with SAROS lol
-----
Download TIGER here! https://tigermeat.xyz/#tiger-ds
Other Projects:
Canary for DiffSinger: https://tigermeat.xyz/#canary-ds
-----
I wuv dis song, i've wanted to cover it since the second i heard it, I may or may not have made this voice mode just because of how crazy good saros is too LOL
USTx: Me (it's not full so i wont release it sowwy)
----------
Hi! Tiger can sing in french now, i'll be releasing this update soon :3
Canary is owned by Mina Moonrise / youtube.com/channel/UCSkSDPVl3W5LWYR2O0OzMUA
Cover/Mix/Vid: tigermeat / twitter.com/tigermeat_
Video Assistance: Guillotama / twitter.com/Guillotamaa | youtube.com/c/Tasteloid
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
Download Canary for DiffSinger here: https://tigermeat.xyz/#canary-ds
Canary's Webpage: http://jaevocalproject.weebly.com/canary.html
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
Today is Canary's 5th Anniversary, happy birthday Canary! Enjoy this cover of "null", a song that I think fit his "Spark" voice mode PERFECTLY!
Canary's voice provider Mina has covered this song like 400 times, so I thought it would be a fitting surprise to have it be one of Canary DS' demos :3c
Canary has VERY little Japanese in his dataset, so this would be the most recent example of my DiffSinger databases Japanese XLS (Cross Lingual Synthesis). The corpus this model was trained on has about 7 hours of data, 30 minutes of native Japanese and about another 30 minutes of non-Native (but proper accent) Japanese. You can certainly tell that Canary isn't native Japanese, but I think the accent is really good for an "English" voicebank!
Canary is owned by Mina Moonrise / youtube.com/channel/UCSkSDPVl3W5LWYR2O0OzMUA
Cover/SVP/Mix/Vid: tigermeat / twitter.com/tigermeat_
Video Assistance: Guillotama / twitter.com/Guillotamaa
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
Download Canary for DiffSinger here: https://tigermeat.xyz/#canary-ds
Canary's Webpage: http://jaevocalproject.weebly.com/canary.html
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
Download the SVP on my website: https://tigermeat.xyz/#seqf
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
Here's Canary DS singing "Make Believe". I've wanted to cover this song since it came out, and Canary sounds really good on it!
This cover utilizes his "Arc" (Standard) and "Voltage" voice modes, while having some of "Spark" used as a curve in the verses!
Canary is owned by Mina Moonrise / youtube.com/channel/UCSkSDPVl3W5LWYR2O0OzMUA
Background Vocals by TIGER DS / https://tigermeat.xyz/#tiger-ds
Cover/Mix/Vid: tigermeat / twitter.com/tigermeat_
USTx by tigermeat & KAISKE / youtube.com/c/KAISKE
Video Assistance: Guillotama / twitter.com/Guillotamaa
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
Download Canary for DiffSinger here: https://tigermeat.xyz/#canary-ds
Canary's Webpage: http://jaevocalproject.weebly.com/canary.html
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
Download the SVP on my website: https://tigermeat.xyz/#seqf
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
This is the 4th and FINAL Canary DS Launch demo! The second of the secret demo's, my boyfriend was like "Canary would sound really good on Naturally" by Selena Gomez... lo and behold, he does sound VERY good!
I had started this UST back in the PINKU ENGLISH 2.0 days! It's old, and the reason I never finished it was because it never synced up right when I mixed. So, I figured out that the tempo of this song is actually REALLY FUCKED UP??? I got the base down but it still felt wrong...
So, I asked KAISKE for a bit of help. KAISKE is an amazing voice synth producer, his cover's are TOP !!!!! TEIR !!!! Go give him a listen! (linked above). He's the master of tempo changes in songs, so he corrected the USTx for me and I am FOREVER grateful.
I hope everyone Enjoyed the canary release! I had so much fun making him and he's not done.... there's more to come....... eventually.
USTx by Gianloop / youtube.com/watch?v=Fn4cAEBkzMk
Canary is owned by Mina Moonrise / youtube.com/channel/UCSkSDPVl3W5LWYR2O0OzMUA
Cover/Mix/Vid/Tuning Edits: tigermeat / twitter.com/tigermeat_
Video Assistance: Guillotama / twitter.com/Guillotamaa
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
Download Canary for DiffSinger here: https://tigermeat.xyz/#canary-ds
Canary's Webpage: http://jaevocalproject.weebly.com/canary.html
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
PADAM PADAM!!!! Canary KILLS this song! This song primarily uses his "Arc " (Standard) and "Voltage" voice modes!
Canary is owned by Mina Moonrise / youtube.com/channel/UCSkSDPVl3W5LWYR2O0OzMUA
USTx/Translyrics: Mina Moonrise
Cover/Mix/Vid/Tuning/Translyrics: tigermeat / twitter.com/tigermeat_
Artwork of Canary: Angela M Chong / ko-fi.com/angelamchong
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
Canary's Webpage: http://jaevocalproject.weebly.com/canary.html
⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆⋆⁺。˚⋆˙‧₊✩₊‧˙⋆˚。⁺⋆
Please enjoy this color coded kpop lyrics video for this beautiful demonstration of canary's voice i made!
He's coming out 2/2!!!! SO EXCITED!!!!!!!
NOTE: THIS IS NOT DIFFSINGER LOL this is the audio files that are used to train tiger diffsinger
Also ignore the weird video, i'm still learning after effects.
Fresh: Basic voice mode, how I sing without putting any effort into making it be specific. This Voice mode has the most non-English recordings.
Disco: I wanted Tiger to have a soft, but not too soft voicemode, like "weak" or "chill", so that's how Disco was born!
Electric: I can sing pretty powerfully, so that was the goal here. It doesn't hit any limits in terms of what I CAN sing, but it's a solid attempt at some power.
Vinyl: My boyfriend has me really into musicals, so I wanted to push my skills and get a more mature sound...
Mystic: This is the more indie, "cursive" singing, but I can't really do that to the extent some people can, so it's a very soft, expressive voice mode.
Falsetto: ........ it's in the title
Resonant: This one I recorded today, as of writing. I wanted to really push my belting, and I was able to get some good samples, so I'm excited for how it'll turn out!!! (Name my change)
---------
Original Song: Captain Mirai
Translyrics: Me based on the original jpn + the official english
Ust: felichanxx
Canary belongs to Mina Moonrise
---------
I've been developing Canary on DiffSinger! He's planned to release on 2/2/24! Keep an eye out for then!
He comes equip with 3 voice modes:
- Arc (Standard)
- Spark (Soft, used in this cover)
- Voltage (Power)
And can also support Japanese and Spanish via XLS!
youtube.com/watch?v=Zx4Hjq6KwO0
Cover: Guillotama of Tasteloid
twitter.com/Guillotamaa
youtube.com/c/Tasteloid
DOWNLOAD TIGER: https://tigermeat.xyz/#tiger-ds
Feel free to ask questions! I'll be releasing the next part soon...ish?
Please note: This guide isn't meant to be professional, so I say silly things, make mistakes and use swear words.
Prerequisites:
- The most basic knowledge of using a computer + python/conda
- Have Anaconda3 installed (anaconda.com/download)
- Have a training-ready DiffSinger dataset
If you do not have a dataset ready, don't fret! Here's a really in depth guide by Julieraptor to get you started!! docs.google.com/document/d/1zn_GZKGAuhPM7np88aOgDH052tAE44U7oLn3O4r090U/edit#heading=h.mnlnkbi32wrd
Another guide by Pix and SpoopyAce: docs.google.com/document/d/1uMsepxbdUW65PfIWL1pt2OM6ZKa5ybTTJOpZ733Ht6s/view
What we cover in this part:
- Setting up the local environment with Conda + Git
- Manually editing the acoustic config file
- Binarizing and Training
- Reading + Understanding Tensorboard
I'm really active in the DiffSinger discord server! There's tons of support there as well, feel free to join! discord.gg/MVaqkwbnT5
Links (kinda in order):
DiffSinger Repo: github.com/openvpi/DiffSinger
nsf_hifigan from openVpi: github.com/openvpi/vocoders/releases/tag/nsf-hifigan-v1
rmvpe model: drive.google.com/file/d/1TTrt2GKn4ZaIbdSuJVgS18aqWz9IsSSw/view?usp=sharing
torch command: pytorch.org/get-started/locally
nnsvs-db-converter: github.com/UtaUtaUtau/nnsvs-db-converter
commands used:
Anything inside of curly brackets ({ }) will be a variable that YOU need to change.
install git: conda install -c anaconda git
create environment: conda create -n {name_of_env} python=3.8
activate env: conda activate {name_of_env}
clone diffsinger repo: git clone github.com/openvpi/DiffSinger.git
binarization: python scripts/binarize.py --config {path_2_config.yaml}
training: python scripts/train.py --config {path_2_config.yaml} --exp_name {name_of_experiment} --reset
Chapters: (Contributed by @ssharkteeth )
00:00 Intro and Requirements
01:52 Setting Up Conda Environment
03:21 Cloning Diffsinger Repository
04:09 Installing nsf_hifigan and rmvpe
05:26 Installing Torch and Diffsinger Requirements
07:08 Data Folder Setup
09:09 Setting Up The Config.yaml
20:54 Binarization / Preprocessing
23:15 Begin Training (use ctrl+c to stop training)
25:54 Tensorboard Setup and Explanations
32:00 Outro
============
Original Song: THE Dixie Flatline (youtu.be/8qCmmGxs5Zk?si=EqCI0pfcF8ByBZ1b)
(Note: The readme in the download says there's no instrumental... there is, i'm just an idiot)
unedited svp download: drive.google.com/file/d/1Xvldfv92CBCnfLTnzcIshweikuTRAOAe/view?usp=drive_link
edited svp+mix flp is on my Ko-Fi! ko-fi.com/s/df0e518c45
============
*SLAMS FISTS ON TABLE* GO LISTEN TO THIS SONG NOW!!!! It's been out for 2 weeks and only has like 5k views???? DO NOT LET THIS SONG FLOP!!!!
I've been meaning to cover this since it came out, and i wanted to see 1) how good SAROS' japanese was and also how femme I could get them to sound without it sounding too forced, both of these things worked very well.
I'm having a lot of fun with SAROS.
SAROS is fucking INSAAAAANE. Literally could not have asked for a better voice period. Go buy them rn!
This is a test of the "audio to midi" function in SV 1.11! I had so much fun tuning this. Stopped tuning at "I'll get you a satanic mechanic"
Lay All Your Love On Me by ABBA | SVP by KAISKE
Clocks by Coldplay | VSQx by wintercoffee
Butterfly on your Left Shoulder by Nori-P | UST by RewoundClockwork
--------------------------
Link to download/install guide: github.com/spicytigermeat/tiger_diffsinger/wiki/Installation
Link to TIGER's website: https://tigermeat.xyz/#tiger-ds
Link to TIGER's Terms of Usage: https://tigermeat.xyz/#tiger-tou
--------------------------
TIGER is HERE!!!!!! I've been working on Tiger DS for quite some time, and I'm finally ECSTATIC to release him! You'll find all the information you need to install, use, etc him from the link above! Below, I'll point out some details about these demo's and explain what will happen in future TIGER releases, and where you can keep your eye out for updates!
Please ensure you read the guide in the link in the previous section to ensure you've correctly installed TIGER.
Notes:
- The cover of "Clocks" in this was not manually tuned, but instead was just the built-in pitch model based on TIGER's voice. Small edits were made during mixing, but nothing in editor.
- TIGER currently supports English mainly, but can sing in Japanese and approximate more languages. Guides are on his Github page!
Install TIGER for DiffSinger: github.com/spicytigermeat/tiger_diffsinger/wiki/Installation (links are in this article!)
-=-=-=-=-=-=-=-=-=-=-
He's back! If you haven't heard, Alter/Ego BONES will be receiving an update soon, and a lot of progress has already been made! I've wanted to update BONES for quite some time, but with all the most current voice changing tools and new ways to upgrade old, outdated audio, it felt like the perfect time! The developer of Alter/Ego LEORA, Infoholic, and I have been rebuilding BONES from the ground up using ONLY his original samples and other voice modifying software (RVC, etc.)
Here's some technical information for those curious.
- BONES-C uses the original recordings as a base, and expands upon them with modern technology.
- Instead of 2 voice libraries, "en" and "jp", BONES-C will be consolidated into a single voice library.
- BONES-C will include a "Soft" slider, which you can hear being used in this sample.
- BONES-C is currently in an alpha stage, meaning the voice db and visuals are subject to change!
- BONES-C is going to be completely reconfigured by Infoholic and I to better utilize the Formant Singer 2 engine via Alter/Ego to provide the best quality results on the engine, while not taking up too much RAM in usage. Currently, the total DB size is roughly 250MB, which is slightly more than BONES' original English DB, but about half the size of other Alter/Ego voice libraries like Marie Ork Clean.
This project is going to be somewhat narrative driven, below is a quick synopsis of the backstory to BONES' return!
BONES -CONSTRUCT-, or BONES-C for short, is a wandering witch-doctor who travels in the wastelands of a desolate and barren futuristic wasteland. He hopes to help society rebuild itself and yearns to find the answers about who he really is.
I hope you're as excited about this project as I am!!!
ustx: me bitch
--------------------
Tiger DiffSinger is exist! Lots of work still left to do, but here's a vocal-focused mix of examples of his vocal modes! I'll write a little bit of info on each below:
---
FRESH | "Natural"
The basic vocal mode, with the most data carrying it. Solid and bright tone, definitely the most dynamic vocal mode that can fit a lot of places.
Recommended Genre: Any, yo.
---
DISCO | "Soft"
This vocal mode is a softer, but not whispery vocal tone. You'll find the high range has more falsetto, and it will fit songs that require less oomph.
Recommended Genre: Disco, Ballad, Indie, etc.
---
ELECTRIC | "Power"
This vocal mode is referencing songs with a lot more impact. This mode has a clear and powerful tone, and won't hit as much falsetto as the others.
Recommended Genre: Rock, Pop
---
VINYL | "Mature? Theatrical?"
This one was... an experiment LOL. The data that this mode references is from musicals and songs that sound like they should be in musicals, but I tried getting my voice more deep and mature.
Recommended Genre: Musical, Showtune-y
------
Some Info about this DiffSinger:
- Supports the basic arpabet set of phonemes
- Extra Phoneme's include:
[dx] - American tapped r, "Butter" [b ah dx er]
[ax] - Proper Schwa, "Again" [ax g eh n]
[vf] - Vocal Fry! You can hear this in the demo :3
[q] - Glottal Stop, "Ointment" [oy n q m ax n t]
You can use [q] after a [t], [k] or [p] at the end of a vowel to make it a stop, very useful for [t]!
I'll provide some more updates here as I have them, but I'm not forcing myself to adhere to a strict date, but I will be releasing this VB! If you want to see more, follow me on twitter/x at twitter.com/tigermeat_
Instrumental/Tuning/Mix/Master: Me
Hi Friends! I've been hard at work developing a text-to-speech GUI application the last while and wanted to share some progress on here. Posting videos on here is pretty out of the way so for regular updates, you'll want to follow my twitter account which is @/neutrogic without the /
I have quite a few goals with this software but 1: It will be free and 2: it will be open source upon release. It uses "TalkNet+HiFi-GAN" as the synthesis engine and "WORLD" as a vocoder for voice parameter editing!
Please keep an eye out for updates and all that jazz! I'm having a blast and have a lot of plans for this project!
Also, I just revealed Tiger's design over on twitter! twitter.com/NEUTROGIC/status/1519722881688735748?s=20&t=A9a_-wDJQCNB0sdJySezTQ
Want the tuned SVP and want to support my work? Purchase it here! ko-fi.com/s/afd770a512
HI I love this movie and have wanted to cover this song forever.
SOLARIA is absolutely astounding, and she can sing just about anything you throw at her.
I referenced my own singing for the tuning on this, but it was still just a silly little thing I made.
Here's who is singing who's part:
Janet - SOLARIA
Brad - Saki AI
Rocky - Qing Su
Dr. Frank N Furter - Tsurumaki Maki ENG AI
Magenta - Eleanor Forte AI
Riff Raff - Koharu Rikka AI
Columbia - ANRI
No Copyright Infringement Intended!
Want the fully tuned SVP? Find it on my Ko-Fi! ko-fi.com/s/9fbe30d2c7
Y'ALL should watch centaurworld, it's a really good show.
i figured out an easy way to throw together quicker videos for short covers like this so I might be posting some more stuff here! I want to put less spotlight on the covers I make vs the arrangements/originals I make so I think this might be my new format, who knows!
Vocal: Synthesizer V Eleanor Forte AI by VOLOR
Vocal and Audio Manipulation / Remix / Artwork / Video: NEUTROGIC
INSTRUMENTAL: drive.google.com/file/d/14-s9h3-AxK44jtemfxZZ8oN0Ifj6KCmM/view?usp=sharing
Hi Friends! I'm really happy that I am finally releasing this! It's been in the works for quite a bit, and every single one of my projects got halted because I got a real nasty case of Covid (I am perfectly healthy now though!)
I'm really happy I could make such a cool cover of a good friend's song, Echo is a banger and I hope I did it justice with this remix. Eleanor might not have been the smartest choice for vocals on this one, but she's all I got for now in terms of really HQ English vocals (IA AI doesn't do it for me)
Please enjoy!
NO COPYRIGHT INFRINGEMENT INTENDED! I CREATED THE INSTRUMENTAL!
Original Song: Lady Gaga
Cover / Produced by: NEUTROGIC (me)
Mastered by: GuitarHeroPianoZero (YT: bit.ly/3oDedUO)
Artwork by: Eggtan (YT: bit.ly/3qYysPV | IG: bit.ly/3DqSMfX)
Hi Friends! I was asked by VOLOR to create a cover for Eleanor Forte's new AI voice database as a demonstration. She's absolutely AMAZING! I can't wait for everyone to be able to try her out!
Thank you to GHPZ for mixing and mastering this and Eggtan for drawing some BEAUTIFUL artwork of Eleanor! Make sure you give them both some love!
Thank you for watching!
KAFU by kamitsubaki
-=-=-=-=-=-=-=-=-=-=-
unedited svp/ccs download: drive.google.com/file/d/1a7PCuQy7M1i2aTqw2UKRc8ravYz9ezG1/view
-=-=-=-=-=-=-=-=-=-=-
aaaa this song is so good!!! And i'm getting the hang of using KAFU too!!! enjoy :D
ust: GMF159
note: i bought kir*tan before all of the NFT and really gross artwork stuff went down with her company, and i do not support that side of it at all. i just like this cover and don't wanna take it down, hence why i'm removing her name from it
-=-=-=-=-=-=-=-=-
tuning n shit: me
-=-=-=-=-=-=-=-=-
this lil song makes me so happy, I've known about it for a while and it's still pretty unknown, at least in the west. Kir*tan's voice fit the song perfectly, so I decided to make the cover hehe.
lq video cover uploaded to youtube for the sake of nostalgia
ust: 地方老人さん (sm36219872)
Saki AI is a blessing from Dreamtonics
tuning/mixing: me
-=-=-=-=-=-=-=-=-=-=-
Happy New Year! I'm still working on my ablum but i've been VERY interested in all of the new AI synth stuff, and I've liked this song for a bit and been meaning to cover it. Saki AI is great, and I'm glad I actually have a reason to use her now, her standard voicebank sounds terrible if you're comparing it to her AI voicebank. It's a very stark difference! Please enjoy, and have a happy and safe new year!
-=-=-=-=-=-=-=-=-
original song+lyrics+stems: 100 gecs
arrangement+tuning: me
midi: half a head/zoe
instrumental download: drive.google.com/file/d/1ItXoZF8-nP5kTmhdVklLu0yDKKc8JYlc/view?usp=sharing
-=-=-=-=-=-=-=-=-
Original PV art: PRiNZ
Vocals: Luka V2
Cover/VSQx/Mixing: me
-=-=-=-=-=-=-=-
vsq/x dl: drive.google.com/file/d/1b7tD0u2c3_FaverOgmLKLkdRv4CCkJq5/view?usp=sharing
-=-=-=-=-=-=-=-
what's up weirdos
i've been obsessed with this song ever since my friend mina showed it to me. It's a BOP and so is all of Lauren's music tbh. It's their birthday (on the day I posted this) so happy birthday!!!!!
original song/オリジナル曲: @x0o0x_
art/絵: 游@stdio_nameraka
-=-=-=-=-
Sequence files: drive.google.com/file/d/1rAZbUhOswF0V5f7aPSycYZFafyhxxuIt/view
(Includes CCS, DV, MIDI, PPSF, SVP, UST, VSQ, and VSQx)
** note that I did this in the original key, the available instrumental is not in the same key, so you need to manually change the key.
-=-=-=-=-
i've internally dubbed this song as saruyume which is what it's based off of, according to some translations. I really like this artists music...
svp, tuning, mix: me (if anyone wants a midi i can upload it, it don't got lyrics doe)
lyricless midi dl: drive.google.com/file/d/1AsIofGQF20Wt9DTkGNMFDul2Eujamwgk/view?usp=sharing
tuned svp: ko-fi.com/s/b4ef6df82e
Haiyi by Quadimension
-=-=-=-=-=-=-=-
YAAAS synth v pro is so good. It's got a few new weird engine quirks, but all in all the sound is wonderful and it's SO easy to use. Like i had SO much fun tuning this!!!! VOCALOID you're fucking dead lmao
I've had Haiyi since basically she came out, and i've wanted to cover this song with her for a while, but i honestly hated using SVR1, SVPro is so much better. I literally love everything about it? It's gonna go places, that's for sure!!!
ust/drawing/mix/utauloid: me
oto configuration: Ryan (twitter.com/ryankun12)
-=-=-=-=-=-=-=-=-=-
ust/vsqx dl: bit.ly/2B0OHnl
-=-=-=-=-=-=-=-=-=-
okay so this song gets me really emotional???? I made a cover of it with SeeU in 2013 and it was one of my first covers/anything to get decent attention and for some reason the key this song is in (lyrics too) just get me in the feels and the nostalgia
sorry i haven't been paying attention to pinku jpn!!! I just........ care more about english lmao. I had so much fun tuning this though, release will be 6/21!!!!!!!!
vsqx: mira + roki
-=-=-=-=-=-=-=-
This is another singing VOICEROID cover, this time with Kir*tan and Mikoto (gynoid talk)!!!!
I've been having so much fun with VOICEROID, making them sing is a freakin' blast!! I've also been OBSESSED with Hime and Mikoto but I don't really care for their V5 voicebanks, something is just a little off tbh.
I'm definitely not dead, but I don't see myself using VOCALOID a lot at this point. It's just no fun anymore.
My UTAU is getting some big updates, so keep an eye open for demos soon!
**I know there's lyric errors, I just used the lyrics on the VSQ***
UST: 某只泽
Singer: ギャラ子TALK / Galaco TALK
-=-=-=-=-=-=-=-=-
:3c did i trick u
no galaco yellow doesn't exist, but galaco talk does! It's the same process as making a voiceroid sing.... sincei 'ts the same engine. I've always had such an interest in making voiceroids sing and I actually had so much fun doing this!!!
ok pls enjoy this song is a bop
.dv/.vsqx/.ust: soon
=========
Original Song: ToToo
art: catticoon
.dv,vb,cover: me
=========
Hey y'all, so pinku CHN is done!!! I'm glad that, albeit on seperate programs, pinku is trilingual now!
i just got a laptop, so now I"m not tied to my desk when I wanna use a computer! I can't record at my desktop anymore because my fish tanks are loud
keep an eye out for more DV stuff from me, as Pinku will not return to UTAU ever again. periodt.
Enjoy!!


