text to speech with the voices

[pearl mountain]

is there any?

I have a nividia 3050

[uneven vigil]

if you're talking about using TTS with RVC v2 voices, Retrieval-based-Voice-Conversion is STS, not TTS natively

the only way would be using another TTS as the base, with a voice model that is using the same language and gender as the RVC one
Then use that TTS output as RVC Input

the RVC Applio Fork already does this, using Edge TTS as a base, to make things easier

[pearl mountain]

So it’s possible using what app?
Any tutorial to set up and stuff?

[uneven vigil]

https://docs.aihub.gg/rvc/local/applio/

[pearl mountain]

set up and stuff?

[uneven vigil]

it's the ai hub docs' Applio Guide

[pearl mountain]

and i can just install a voice model and add it?

[uneven vigil]

if you mean RVC v2 models, yes, it's an rvc fork (modified version), that uses Edge TTS as a base for doing TTS with RVC models

[pearl mountain]

nerd it down please , ur wayy to smart

[uneven vigil]

yes, it works with #1175430844685484042

and does TTS

just don't expect a too good or emotional TTS

[pearl mountain]

Alright thanks lmao

[pearl mountain]

Yeahh dw , i just need like yeah and shi for trolling my friends

[pearl mountain]

Hey , theres a error or sm

[uneven vigil]

move the applio folder to a path without spaces

be sure that none of the folders where applio is in, has no space

[pearl mountain]

like these folders or

its just in my downloads rn

[uneven vigil]

could you right click the Applio folder then do Copy as Path?

[pearl mountain]

done

[uneven vigil]

can you paste it here?

you can just right click your mouse and paste what you copied

[pearl mountain]

it kinda contains my rl name

which has a space

so how do i move it

where do i move it

[uneven vigil]

yeah that explains, move the folder to C:\

if you don't know how to do that

move it to: in file explorer -> This PC -> OS (Windows version name)

[pearl mountain]

ight

and its gonna take 4 mins

nvm

how do i add it

[uneven vigil]

check https://docs.aihub.gg/rvc/local/applio/#1-upload-voice-model

after you uploaded it via that section, check https://docs.aihub.gg/rvc/local/applio/#tts for TTS

[pearl mountain]

What type does it need? not pth?

nvm

[uneven vigil]

you seem to have confused the audio uploading section with the model one

[pearl mountain]

Alright i downloaded it

do i restart the window

[uneven vigil]

you don't need to restart windows

[pearl mountain]

its still seem to use this audio instead

of the downladed one

[uneven vigil]

yeah that's normal, it needs to firstly use a TTS model to make the TTS normal output

then, it applies the RVC model over the TTS output

it's not directly TTS, it uses a workaround

[pearl mountain]

oh so its gonna sound like ai

bruh

[uneven vigil]

it uses Edge TTS for the normal tts audio, which is high quality (not like google translate if you get what I mean) and multilingual, but not Emotional

[pearl mountain]

does the realtime work

[uneven vigil]

are you looking to use TTS in realtime for calls as a microphone ?

[pearl mountain]

Yup

[uneven vigil]

yeah, you need to do a bit of manual setup with a Virtual Audio Cable, following: https://docs.aihub.gg/tts/realtime-tts/#virtual-audio-cable

[pearl mountain]

i have a audio cable

[uneven vigil]

VAC Lite (Virtual-Audio-Cable by Muzychenko) or VB Audio Cable?

[uneven vigil]

some users in here reported issues with vb audio cable, such as randomly stop working
We usually suggest VAC Lite (Virtual-Audio-Cable by Muzychenko) instead of that for Windows users

[pearl mountain]

is it better?

[uneven vigil]

To not risk any random issues, it would be better to use the other one in the guide I sent

[pearl mountain]

alr well how do i delete the vb one

[uneven vigil]

Settings -> Apps -> Search it -> 3 dots -> Uninstall

btw be aware that you will need to type the words to make it work, so it's not like "instant" since it's TTS, is that what you want ?

there is another non-tts realtime way if you want, depends on your choice if for example you have any issues with talking or idk

[pearl mountain]

iits not there

[uneven vigil]

try searching for just "vb"?

[pearl mountain]

how do i remove the default

[gentle yoke]

now you should reboot after uninstalling vb cable

[silver yew]

Right click on "Speakers" and set it as both default device and default communication device, so the green icon should gone from Line 1.

And then do the same as in "Recording".

[pearl mountain]

@uneven vigil Hey
after doing some reasearch i found this

https://github.com/w-okada/ttsclient

does it work?

[pearl mountain]

?

[uneven vigil]

you need to set your usual device as default instead

it's in the guide

[uneven vigil]

the last change was 7 months ago, and it supports only GPT-SoVITS (TTS) models, not RVC (STS) models, so it won't work with #1175430844685484042 that contains mostly just RVC models

[pearl mountain]

TTS models still sound real or not

[uneven vigil]

GPT-SoVITS is good, but it's a completely different type of model, the great majority is just RVC in #1175430844685484042

so you will be able to use very few models compared to RVC ones

if you need to use RVC models, why don't you just try the tts realtime guide with applio?

[pearl mountain]

It lags for some reason

Like it’s really slow

[uneven vigil]

wdym with lag? you have to type the words, so it's not really "instant"

how much time does it take to generate the tts on applio?

[pearl mountain]

15 secs

And it’s like really low ai quality

[pearl mountain]

Is there a guide on how to use this

Or installation guide

[uneven vigil]

there's no guide about it in the ai hub docs, the max you can do is read https://github.com/w-okada/ttsclient/blob/master/docs_i18n/README_en.md, but be aware that it doesn't get updates since months and i'm not sure which gpt-sovits model type it supports, nor i have ever used the program (like most helpers)

[uneven vigil]

are you doing that while running heavy/many applications?

or is your wifi slow?

[pearl mountain]

Just Minecraft and discord

[pearl mountain]

Uh

I tried to generate a simple line before it took 10 sec to generate hi in like a really low ChatGPT voice

[pearl mountain]

Does it work tho

[uneven vigil]

Not sure if it has any issues

I never used it

[pearl mountain]

Ight Alr

[uneven vigil]

You should try other other (Edge) TTS and RVC models

You could also try running Applio via Cloud (remote good PC) in case your PC power might be busy with heavy application, but this means you need a good wifi connection and also you have limited free GPU time

[pearl mountain]

Can u tell me the best top tier one I could use with my laptop containing these

A 3050 laptop gpu

[uneven vigil]

You are already using applio locally

[pearl mountain]

Ight so it’s the best one I can use

Got it

[uneven vigil]

What I'm saying is, if your PC GPU is busying playing like heavy Minecraft shaders that occupy the GPU power, you could try cloud

Cloud means it's running the program in a remote good pc

So you can game peacefully

But it will require a better wifi connection and you get free limited GPU time

[pearl mountain]

Alr

Also Do u know a app which can help me filter my voice like make it really deep or loud or like a high pitch

[uneven vigil]

that looks like a basic voice filter rather than AI, like voicemod ig?

[pearl mountain]

alr

[uneven vigil]

can this be marked as solved then?

[pearl mountain]

Yeah