tigermeat

tigermeat

@tigermeat.xyz

27♂️🌈🐯 | official voice synth account | https://tigermeat.xyz/ | contact@tigermeat.xyz

279 followers172 following498 posts

tigermeat reposted

My DiffSinger energy is SLOWLY returning.... so very slowly I have been obsessed with this song so I wanted to see how TIGER sounded on it, the answer is. good. This is a test vb i trained last week, it sounds pretty good Paparazzi Murder Party By: FLAVOR FOLEY SVP: vocaloader.app/sequences/2683

My DiffSinger energy is SLOWLY returning.... so very slowly I have been obsessed with this song so I wanted to see how TIGER sounded on it, the answer is. good. This is a test vb i trained last week, it sounds pretty good Paparazzi Murder Party By: FLAVOR FOLEY SVP: vocaloader.app/sequences/2683

the test VB basically is like 90% data I labelled and a lil extra, it only can sing in English, Japanese and Chinese. The idea was to try out fine-tuning, but with the acoustic model it sounded like SHIIIIIIT

Sour 2.0 for UTAU uses an early version of a tool I'm working on called UTACompiler. It's available to test on my Github (mrtigermeat). Not linking because it's very unfinished and untested, use at your own risk!

View full thread

There will also be an option to include dummy audio to further deter audio scrapers, but this part has proven to be very problematic with FRQ files, so I still need to find a way to implement that without degrading the voicebanks quality.

I can't promise when it'll be done cuz full honesty, I'm beyond burnt out with new job & health issues, but I promise I intend on giving it a full release <3

Sour 2.0 for UTAU uses an early version of a tool I'm working on called UTACompiler. It's available to test on my Github (mrtigermeat). Not linking because it's very unfinished and untested, use at your own risk!

UTACompiler reads a voicebanks OTO and cuts out unnecessary audio, limiting file size. It's very helpful on large voicebanks. It also rearranges sounds, which can obfuscate recording styles & deter audio scraping for things like unauthorized RVC models and such.

#DiffSinger Please do NOT contact me about licensing my DiffSinger libraries in your software if your software was created with. or uses AI in its development. Despite the AI nature of DIffSInger, it is fully local & all data is ethically sourced. I do not support AI in software development.

i've gotten enough of these offers to warrant this post... and I hope none of my peers in the DiffSinger dev community do not sell themselves out to AI grifters either. Getting really sick of it.

tigermeat reposted

My sixth attack was a revenge for really cute Aidyn art I got from my friend 🥺 I've never drawn TIGER before but I think his design is SO funky and original. I really love the electric vibes he has and the beard is so cool! TIGER belongs to @tigermeat.xyz #Art #Artfight #ArtFight2026 #teamcomedy

YAAAAAA go watch Arreen STUN right now it's not optional!!!

I've had the amazing opportunity to work with KRO Zin (Dutch TV channel) to make a video about cosplay and selfexpression! I am beyond grateful for the producers and crew and am so excited about the result. (Video is in Dutch, auto translations available!) youtu.be/NqH084LQ14M

I'm working on a voice synth project RN that I'm super excited about but I don't want to start teasing it yet. It's a big departure from what I've been working on the last few years, and I think people will like it :)

I've decided that I'm not going to rerelease TIGER's public corpus. I'm not particularly happy with the quality, and given the misuse and lack of respect for my license agreement, I'd rather not risk it being misused any further. Plus, there are a lot more public English Databases now.

tigermeat@tigermeat.xyz

TIGER 2.0 Public Corpus Update Due to repeated misuse of the dataset, I've elected to remove the dataset from being public. I'll be opening a form to request access to it soon, but for now it will be inaccessible. If you already have the dataset, you may continue to use it, no redistribution.

If you already have it, you may continue to use it, but must follow the license agreement. I will not be providing support for it in scenarios where it's not working as expected. No hard feelings toward anyone, I'm just tired.

TIGER 2.0 Public Corpus Update Due to repeated misuse of the dataset, I've elected to remove the dataset from being public. I'll be opening a form to request access to it soon, but for now it will be inaccessible. If you already have the dataset, you may continue to use it, no redistribution.

And for the sake of the hard work and dedication this community puts into their creations, it's your responsibility to respect the wishes of the creator. Learn how to read license agreements.

Not sure if it's obvious or not but I am kind of deciding that right now i'm in a voice synth creation/development hiatus period. How long? No idea, but I have too much other stuff I want to/need to focus on rn and just don't have the drive to touch diffsinger

Public reminder to everyone that if you use TIGER's SVS dataset you must follow the TOS! You may not: - Redistribute - Release embeddings of TIGER's voice - Train on SVS/SVC systems other than DiffSinger and NNSVS. I will discontinue the dataset if the TOS is broken continually.

Hi y'all! I'm alive just burnt to a crisp from my new job and dealing with some health issues. I'm totally fine, but don't expect anything from me for a good while until I'm back up to snuff

Man, Mikoto studio is ROUGH. I would never release my software in this state... UX/UI is awful, i mean cmon 2 clicks to draw a note? Not to mention, the synthesis is really rough and not something i'd be willing to use... I'm sorry to be a negative nancy but it doesn't even feel beta tested

tigermeat reposted

If you can donate to help Tama fund her HRT, that would be extremely sexy of you!!

This post is unavailable.

i'm curious if anyone has any thoughts about the new transcription UI for LabelMakr 2.0? Doesn't work yet I plan on adding some options for the transcription (like text cleaning, etc.), and i will be turning the file viewer into a treeview instead of a listbox

i'm currently having a hard time getting custom fonts to show up on Linux 😔 works fine with pyglet on windows but i guess linux and pyglet do not work well together lol

tigermeat reposted

On June 3, 2026, we will be celebrating Xiao Meihua's 10th Anniversary as an #UTAU! That's in 2 months! Here's a look at her 10th anniversary design, with a complementary outfit for Xiao Meiye! The outfits were designed by Ruthrefoard_B and illustrated by TSaianda!

My artwork for my cover of Chikemuri Ryoujo Immoral Taste. I am really happy with how it turned out, and I'm usually really self conscious about my artwork! #DiffSinger #UTAU #VOCALOID #art #tigermeatart youtu.be/-dUtZuSfd9w

Okay, well, I guess it's up now. For some reason, even tho I asked it to premiere at 12PM, Youtube decided to publish it at 7:30AM!! THANKS YOUTUBE!!!!! Please see the full cover here, I'm so proud of it :) youtu.be/-dUtZuSfd9w

youtu.beBitter for #DiffSinger | 血煙旅情インモラルテヰスト (Original PV)YouTube video by tigermeat
tigermeat@tigermeat.xyz

New Cover Premiere - My First Original PV! Please join me this Friday, April 3rd at 12:00 PM CST (no countdown) to see the premiere of my first ever original PV! Bitter DS - 血煙旅情インモラルテヰスト (USTx will be included ;3) youtu.be/-dUtZuSfd9w #DiffSinger #Vocaloid #UTAU #VoiceSynth

tigermeat reposted

New Cover Premiere - My First Original PV! Please join me this Friday, April 3rd at 12:00 PM CST (no countdown) to see the premiere of my first ever original PV! Bitter DS - 血煙旅情インモラルテヰスト (USTx will be included ;3) youtu.be/-dUtZuSfd9w #DiffSinger #Vocaloid #UTAU #VoiceSynth

New Cover Premiere - My First Original PV! Please join me this Friday, April 3rd at 12:00 PM CST (no countdown) to see the premiere of my first ever original PV! Bitter DS - 血煙旅情インモラルテヰスト (USTx will be included ;3) youtu.be/-dUtZuSfd9w #DiffSinger #Vocaloid #UTAU #VoiceSynth

PREMIERE IN 10 MINUTES. THE WORLD WILL NEVER BE THE SAME IN 10 MINUTES.

Hibya@hibya.bsky.social

Releasing new #diffsinger voicebank today-ish www.youtube.com/watch?v=xbl9...

i'm working on beginning work on a LabelMakr 2.0 update, since a lot has changed in the auto-labeler sphere of things for DS since the last time I worked on it. Biggest goals - completely rehaul the UI to be a lot friendlier and incentivize users to actually interact with the transcription editor

- streamline the SOFA support - add support for WFL and HFA - model catalogue/downloader i have no promises of getting this ready any time soon since i start my new job next week, but if you wanna help with contributing code or anything pls let me know :) it will still run on python

As of right now (most) TGM cast members are available on UFR!!! Thank you all once again for this awesome opportunity to collab! 🥳🥳🥳

Voici les dernières sorties du modèle #Diffsinger. Hitsune Kumi, les versions UFR des voicebanks TGM ainsi que Saiun peuvent être téléchargées sur UFR. (1/2)