anons i found a training guide but im stuck.
https://docs.google.com/document/d/1uMsepxbdUW65PfIWL1pt2OM6ZKa5ybTTJOpZ733Ht6s
i was able to create the model but the only characters that work are AEIOU and they sound like a demon being summoned.

Recently found out there's a wiki for Diffsinger and it links a bunch of useful resources, go check it out its pretty cool
https://diffsinger.miraheze.org/

hard to get a diffsinger thread going when only 10% of the community knows it exists. hopefully menhera will give it more recognition if she ever makes it past the concept stage.

Does anyone have Ashera Lyre's voicebank? I tried to get it from OP's voicebank link but the website for Ashera's VB is down.

>>2052
She's discontinued, good luck finding an archived version because no one bothered to archive the page on the wayback machine

c3995b33fc….jpg - 550.10 KB (1080x1080)

>>2053
Forgive me i just found out she's part of the Lunai Project and that site IS archived so here you go anon.

https://mega.nz/folder/LSZG0JYQ#ppsXEDgRCdAqXprW2NmMqA

Are there any English video tutorials on making DS banks? The first one is too vague

neeteto.gif - 127.51 KB (210x226)

You guys are our only hope right now, we at the "fake a SV" thread need you >>1270

Bumping just to ask if anybody knows a decent colab for training

Bump. I need a good tutorial on making a voicebank, preferably with detailed steps.

Does no one here know how to use diffsinger?

>>6659
I know I don’t
I’ll learn when TORIKO DS releases

Once again, we might need you guys

>>7773
Anon... no one here knows how to make DS banks. Why do you think this thread is so dead compared to the others?.

>>7774
The thing with DiffSinger is that a lot of tutorials and information are gatekept on dicksword servers. People here either don't use dicksword or don't like joining public servers.

>>10028
There's a guide linked ITT but it's kind of a pain in the ass to follow

So now that we have 1 (one) person who knows how to use the damn thing can we revive the thread?

Are we ever getting another LEEKA on DS?

>>14237
Maybe Sena next year. A while back I made a RVC of her but it lacked data so it sounded off.

Does anybody have an easy to understand tutorial on how this thing works and all the tools I need? The linked ones are all pretty vague + likely outdated and I'm not joining a dicksword server just to learn how to train a voicebank that I won't even release in the first place

>>15252
Also if we can get a Rentry that'd be nice

>>15252
Seconding

>>15252
there isnt. in order to make toriko i had to have someone walk me through the process step by step over text.

>>15269
sounds tedious

>>15270
it is. if anyone needs help, i can try, but only if youre using difftrainer for local training. even then, my knowledge is limited. if you dont have a beefy computer, im sorry.

How much would you have to record to have a DiffSinger that sounds close in quality to some of the biggest AI synths? I heard that ACE Studio recommends at least 30-100 minutes of audio and while DiffSinger isn't ACE I wonder if it still applies.

>>22579
One of the guides I read said past 2 hours of data doesn't really do much anymore and can even worsen the quality, so I guess that's the cap

>>22582
60m is already 1 hour and 100m is 1 hour with 40 minutes so anything beyond that is probably too much.

>>15284
What amount/GB of vram do you need for it?

>>22579
Synth V supposedly uses 50m

>>2154
Look how far we've come from the days of begging the DS support thread for help, lol

>>22584
I would also like to know because my other option if I don't have enough is to pay Google and train via Colab

>>22597
I think VIP is still the only person here who knows how to make a DiffSinger

>>22584
>>22598
ive fielded the question about vram to someone who knows. if you are unable to train your diffsinger for whatever reason, i can help you out. i will try to write and/or record a guide for using difftrainer at some point in the future. i dont know when, it might take a while. i am a slow worker.

>>22648
It's okay! Take as long as you need! As long as we eventually get one it would be super cool! Even if it took like. a year or more lmao

you need at least 6 gb of vram, and to set the batch size lower to train. the biggest bottleneck for a lot of people is ram apparently. i myself have 16 gigs and have had no problem training toriko, save for user error. the real thing you need to worry about is nvidia cuda compatibilty. if you dont have cuda cores, youre fucked and have to train it via colab or ask someone else to train it locally themselves for you.

>>22676
I have 4gb vram so fuck me I guess. Colab is my only hope.

>>24518
are you ena's vp?

>>24537
No

I saw that some pages recommending that sample songs don’t exceed 140bpm, probably to make sure there’s sufficient data in each phoneme. Does this hurt the model’s ability to sing fast songs though?

If I remember correctly TORIKO’s 1.1 update improved her ability to sing fast, and I wonder if that was due to increased high-BPM data or improved labeling.

How long does it generally take you guys to render shit in DS?

>>24606
Don't know about 1.1 but 1.2 can sing fast songs for the most part even if they exceed 140BPM.

Hello, I am stupid, once you download a vocoder how do you set it in the diffsinger's dsconfig.yaml? do I just put the file name (and extension)?

please ELI5 I have 2 brain cells

>>25584
Open the config and if it should say something like nsf_hifigan and you just replace that with the name of the oudep you downloaded

>>25584
open it in a text editor and just change the vocoder listed there

>>25598
without the extension btw

I found what seems to be an auto labeler and a label checker.
https://github.com/spicytigermeat/labelmakr
https://github.com/spicytigermeat/labbu
Anyone tried these?

>>25949
They make a label checker and name it Labubu, gg

>>25949
The checker looks helpful. Maybe auto label generation is helpful to start, but I’d heavily verify the output. I know moresampler’s CV auto-oto is wonky and it should be using the same techniques.

>>25949
wish there were instructions on how to use it, this and difftrainer

>>25950
Labbu*

I found a RVC->DiffSinger tool which might speed up the process of making voicebanks but at the same time I tested it on an RVC I had made and I ended up with a completely different voicebank that sounded nothing like mine so idk.
https://github.com/SaigyoYuyuko/RVC-Packer-for-OpenUTAU

>>26389
Just tried it and results are very hit or miss. Might be a good idea to stick to manual labeling via vLabeler.

adamsan.mp3 - 2404.14 KB

>>26389
adamsan is finally real no way

>>26409
Nice to see people still remember adamsan

>>26410
>Adamsan
I need context

>>26412
A joke fanloid made on this board that was made after someone mistook that old Leon fan-design for Adam Sandler

>>26389
>>26409
The Vocaloid Piracy people on Reddit are gonna have a field day with this shit. Expect Teto SV on DiffSinger along with KAFU and Spongebob because you just know someone's gonna use it to port Spongebob.

>>26412
Here: >>4668

>>26421
eh, even if i don't like reddit i gotta respect the autism of the vocatorrent people. their stuff gets taken down all the time and they reupload it a thousand more times.

>>26421
eh, people from the diffsinger's dicksword server are pretty upset about it, but imo it's not like you couldn't train an illegal diffsinger before, and i'm sure people did. this just makes it easier

>>26428
Non-RVC DiffSingers sound better anyway. These sound way too noisy.

>>26421
there's a thread here for ports and stuff
>>1

Genuinely though why is it so hard to find good DiffSinger guides

>>26624
You can thank dicksword for that.

>>26624
if its any consolation, im working on one. it should be done at the same time that sena DS is done

>>26633
doing god's work

>>26625
dicksword was the worst thing to ever happen to UGC and modding communities

Is there a way to make Diffsinger render faster? I tried to make my computer run Toriko and it was like she rendered infinitely. I have 64GB of RAM and 8GB of VRAM so unless you need CUDA cores to run a Diffsinger (even if you're not training) that confuses me.

>>26642
How long does it take for you? I have 8GB RAM and 4GB VRAM which is less than you have and it renders in about 10-20 mins depending on how many tracks there are in my ustx.

>>26633
Where will you post it?

>>26644
youll see. i'll link it when its done

>>26642
Is DS really that resource intensive, i have a low spec pc and it works fine. Takes quite a bit to render but not literal hours.

>>26650
I don't have problems rendering either, but I remember one of the repositories for Diffsinger says to only train it on a machine that you're not scared of losing lol. I'm kind of scared to try because of that warning

>>26657
You'll be fine on Colab but you'll probably have to train because Google limits how much you can use the free versions for

>>26657
Depends on your specs

A laptop from 2005 would probably kill itself but a modern PC would be able to handle it

>>26643
It just says 'Waiting Rendering.' No error, no sign it's actually rendering. I wonder if I'm missing some sort of dependency to run DS since it no longer runs any of mine and I had to move my OU folder to my C drive, when previously it'd take 15-30 minutes to render something depending on the length.

>>26672
Check the log files, but I think if you were missing a dependency it would just throw an error instead of hanging.

>>26672
Did you try Lunai or UtauV fork? Fixed most of mine issues

>>26672
You have the Vocoders installed right?

>>26673
>>26678
>>26679
I got her working, it was a vocoder issue unsurprisingly. The only reason I didn't expect it is because I moved the folder rather than reinstalled

>>26723
Reminder to swap them because the default vocoders kinda suck

>>26724
I'm pretty new to Diffsinger, how do I do that and do you have any recommendations? I've only really worked with normal UTAUs in OU and V4 for long periods, so this is sort of new even if the editor is familiar

>>26726
NTA but you gotta install the new vocoders manually and then change it in TORIKO's text file. Vox people should modify the vocoders so we don't end up downloading the version of her that sounds like she has a boulder stuck in her throat.

>>26728
Maybe it would be ideal to have the recommended vocoder in the .yaml file and link to the download for it from TCC. I still think the vocoder should be set per project like a resampler.

>>26732
I agree. Don't know why they haven't done this yet.

I changed the vocoders and I don't hear the difference... Did I do something wrong? That's how it's in the file:
vocoder: kouon_nsf-hifigan_1031
Same with Toriko

ClipboardI….png - 45.23 KB (277x554)

>>26942
These are my settings. Did you install the vocoder from the oudep?.

ritsu_befo….mp3 - 592.53 KB

>>26943
yeah, dragged and dropped them into utau window. that's how my toriko's file looks too, and for Ritsu I tried 1031.
Warning for loud btw. I really don't think it did anything.

>>26944
Delete your UTAU's cache and restart, and if that doesn't fix it then idk

>>26944
What fork are you running this on?

>>27177
lunai

I know this is the support thread but can somebody explain the difference between DiffSinger and NNSVS/ENUNU and which one I should choose if i want to make a voicebank?

by the way, i wrote a tutorial for making a japanese diffsinger and training locally using difftrainer. if anything is poorly written or could be explained more concisely, let me know. i will write a guide for using diffsingers in the future, but right now my main priority is fleshing out the creation guide, since a lot of it is intuitive to using openutau in general.
https://vipchansdiffguide.neocities.org/

>>27507
Different technologies plus DiffSinger sounds a lot better than Enunu. Here's Ritsu's Enunu and DiffSinger voicebanks for comparison.

Enunu:

https://youtube.com/watch?v=nXSC1RANRrE [Embed]

DiffSinger:

https://youtube.com/watch?v=D0pTiop25PU [Embed]

>>27759
Thank you vip

>>27760
Enunu sounds extremely noisy

>>27759
I have an AMD card from the 2010s, fuck

>>27831
Same here but i believe there's a colab file out there

>>27831
if, for whatever reason you cannot use colab, you can send the files and i can train it for you. i used to also use an amd card from the (early) 2010s, so i understand your plight and am happy to lend my hardware.

I wish Tunelab had Diffsinger support

>>27865
OpenUTAU is literally the same thing

>>27844
Nta but link the colab

>>27915
Also consider making a rentry instead of a neocities

>>27915
i dont know what the correct colab link is, i never got it to work properly

if anyone needs help labelling i'm avaivable

A bit unrelated to DiffSinger but some here might find this interesting

https://x.com/canon_73/status/2083495242150649966 [Embed]

Show the video in this Twitter/X post

Canon made a tutorial on training ENUNU using cloud GPU

i've been thinking about how diffsinger doesn't have its own dedicate vocalsynth and instead relies on OU/TLQ. why does ENUNU have one but diffsinger which is objectively better and closer to synthv in quality doesn't?.

>>32067
Since when does ENUNU have its own software

>>32070
Canon made it,lol

>>26944
my ritsucopter goes la la la la la la la la la la la la la la la la la la la LA LA LA LA LA LA LA LA LA LA LA LA LA LA LA LA SOI SOI SOI SOI SOI SOI SOI SOI SOI SOI SOI

>>32067
That's DiffScope and it's still in development

>>32103
First time i'm hearing about it. I hope it includes a feature to switch the vocoder directly from the software without having to do it manually.

>>32103
>DiffScope
Tell me more

Can someone recommend me an OpenUTAU fork that works with Tori? Lunai's works but the recent versions aren't the best for regular UTAU stuff and I can't even install oudeps by dragging and dropping anymore. UtauV is perfect except for the fact that Tori's Diffsinger voicebank is completely silent when i use it.

>>36894
Is it just Tori or other Diffsingers do that too? Did you try swapping her vocoder?
I'm afraid your option is either tunelabQ or vanilla OU

>>36894
worth asking but did you change the phenomizer? toriko's default is DIFFS instead of either the EN or JP diffsinger phenomizer