Software Engineering Courses (SECourses)•2y ago

Stage 1 now works fine. But in stage 2 I have this error: "Input type (torch.cuda.FloatTensor) and w

Stage 1 now works fine. But in stage 2 I have this error: "Input type (torch.cuda.FloatTensor) and weight type (torch.FloatTensor) should be the same"

JJaviEM Stage 1 now works fine. But in stage 2 I have this error: "Input type (torch.cud...

Furkan Gözükara SECourses•3/5/24, 12:46 PM

good luck to fix :d

JaviEMOP•3/5/24, 12:46 PM

JaviEMOP•3/5/24, 12:52 PM

found this on internet: "suggests that your inputs are on the GPU, but your model is not. So I’d double check whether your model is effectively on the GPU."

banfi6885•3/6/24, 7:41 PM

Hello @Furkan Gözükara SECourses, which local text-to-speech model do you recommend for longform content that is comparable to ElevenLabs (without the price tag)? Thank you

banfi6885•3/6/24, 8:44 PM

@Furkan Gözükara SECourses, BTW, I watched your Master Deep Voice Cloning in Minutes on YouTube.

Bbanfi6885 Hello <@205854764540362752>, which local text-to-speech model do you recommend f...

Furkan Gözükara SECourses•3/7/24, 2:01 AM

cocuqi is best i think

Bbanfi6885 <@205854764540362752>, BTW, I watched your Master Deep Voice Cloning in Minutes ...

Furkan Gözükara SECourses•3/7/24, 2:01 AM

i plan to make ane updated tutorial with coqui hopefulyl

Dazzastrous [4090]•3/7/24, 2:01 AM

nah i think tortoise is better

Dazzastrous [4090]•3/7/24, 2:02 AM

rvc

DDazzastrous [4090]rvc

Furkan Gözükara SECourses•3/7/24, 2:03 AM

rvc cant make text to speech as far as i know

DDazzastrous [4090]nah i think tortoise is better

Furkan Gözükara SECourses•3/7/24, 2:03 AM

tortoise was ultra slow compared to cocuqi last time i used

FFurkan Gözükara SECourses rvc cant make text to speech as far as i know

Dazzastrous [4090]•3/7/24, 2:04 AM

yep it can i have rvc tts

Dazzastrous [4090]•3/7/24, 2:05 AM

better quality than Coqui.....Coqui wasnt very accurate for me

Dazzastrous [4090]•3/7/24, 2:06 AM

too american sounding

DDazzastrous [4090]yep it can i have rvc tts

Furkan Gözükara SECourses•3/7/24, 2:11 AM

probably it is using something else to generate speech

Furkan Gözükara SECourses•3/7/24, 2:11 AM

and then apply rvc

Dazzastrous [4090]•3/7/24, 2:12 AM

can make the voices sing too

Dazzastrous [4090]•3/7/24, 2:13 AM

https://github.com/litagin02/rvc-tts-webui

GitHub

GitHub - litagin02/rvc-tts-webui: Text-to-Speech Gradio webui using...

Text-to-Speech Gradio webui using RVC and edge-tts - litagin02/rvc-tts-webui

DDazzastrous [4090]can make the voices sing too

Furkan Gözükara SECourses•3/7/24, 2:13 AM

to improve quality are you using rvc training?

DDazzastrous [4090]can make the voices sing too

Furkan Gözükara SECourses•3/7/24, 2:13 AM

yes it uses edge-tts

Furkan Gözükara SECourses•3/7/24, 2:13 AM

to generate speech

Furkan Gözükara SECourses•3/7/24, 2:14 AM

https://github.com/rany2/edge-tts

GitHub

GitHub - rany2/edge-tts: Use Microsoft Edge's online text-to-speech...

Use Microsoft Edge's online text-to-speech service from Python WITHOUT needing Microsoft Edge or Windows or an API key - rany2/edge-tts

Furkan Gözükara SECourses•3/7/24, 2:14 AM

haha clever using remote service of microsoft

Furkan Gözükara SECourses•3/7/24, 2:14 AM

edge text to speech is really high quality

FFurkan Gözükara SECourses i plan to make ane updated tutorial with coqui hopefulyl

banfi6885•3/7/24, 2:15 AM

Great. Thank you.

andrew.beniston•3/7/24, 4:31 PM

Hey all, having great fun with SUPIR V22 right now - once in a while (havent figured out why) I got an error with the face restoration, I thought maybe its something with resolution but I will break on a 1920x1080 image or others too. Here are the errors, any help is appreciated!

Aandrew.beniston Hey all, having great fun with SUPIR V22 right now - once in a while (havent fig...

JaviEMOP•3/7/24, 5:23 PM

Write a prompt on "face prompt" box

andrew.beniston•3/7/24, 5:29 PM

I literally just figured it out, thanks for that too though!

Aandrew.beniston I literally just figured it out, thanks for that too though!

Furkan Gözükara SECourses•3/7/24, 10:10 PM

it is supposed to use same prompt but i will investigate

FFurkan Gözükara SECourses it is supposed to use same prompt but i will investigate

andrew.beniston•3/7/24, 10:17 PM

Roger that!

Zono50•3/8/24, 3:39 PM

@Furkan Gözükara SECourses What's the best way to use open source to transcribe a 2 hour video on my local machine?

ZZono50 <@205854764540362752> What's the best way to use open source to transcribe a 2 h...

andrew.beniston•3/8/24, 4:02 PM

I would use I would use https://github.com/jhj0517/Whisper-WebUI this comes with Open AIs whisper built in. Take your 2 hour video transcode to mp3 then run it through, it'll take a while but it'll be accurate!

GitHub

GitHub - jhj0517/Whisper-WebUI: A Web UI for easy subtitle using wh...

A Web UI for easy subtitle using whisper model. Contribute to jhj0517/Whisper-WebUI development by creating an account on GitHub.

Zono50•3/8/24, 4:03 PM

Thank you

andrew.beniston•3/8/24, 4:03 PM

No problem, shout if you get any errors!

Zono50•3/8/24, 4:03 PM

does this only do subtitlies oor can i export all the words to a text file

andrew.beniston•3/8/24, 4:04 PM

It will literally spit all the words in to a textbox

Zono50•3/8/24, 4:04 PM

awesome

andrew.beniston•3/8/24, 4:04 PM

I use it for all my meetings with clients, does the best job at transcribing I've seen so far since I use Google Recordings too

Zono50•3/8/24, 4:04 PM

i'm assuming this works on linux

Zono50•3/8/24, 4:05 PM

yup, i see a .sh file, perfect

andrew.beniston•3/8/24, 4:05 PM

As long you have your requirements CUDAs etc

Zono50•3/8/24, 4:12 PM

what's the best way to install cudatoolkit in conda?

Zono50•3/8/24, 4:12 PM

actualyl nevermind my system is already on 12.4, i should be good

bgtubber•3/9/24, 1:41 AM

Hello, guys. New member here.

I was eager to try Supir locally and found this community. I decided to join the Discord and the Patreon. Unfortunately, I'm getting an error so I hope some of you smart and kind people might know what's going on. I downloaded "SUPIR_v22.zip"from the Patreon page and ran the "windows_install.bat" file. After a long wait for things to finish installing, I ran "direct_start_no_options.bat". Immediately after running that BAT file, I'm getting the following error (please check screenshot attached). I'm using Python 3.10.11, (I also tried 3.10.9), Windows 10 x64 Pro with the latest updates. My hardware is as follows: Ryzen 7950x 16-core 5.7 Ghz, 64GB DDR5-6400, GeForce RTX 3090 24GB. Any ideas what could be the problem? Thank you so much in advance!

bgtubber•3/9/24, 1:45 AM

Here's the install log as well in case this gives some insights.

supir_windows_install_log.txt13.93KB

Aandrew.beniston I would use I would use https://github.com/jhj0517/Whisper-WebUI this comes with...

Zono50•3/9/24, 1:46 AM

i'm getting a weird error when i try to grab text from this 2 hour video - /home/zono50/anaconda3/envs/whisper/lib/python3.10/site-packages/torch/cuda/memory.py:303: FutureWarning: torch.cuda.reset_max_memory_allocated now calls torch.cuda.reset_peak_memory_stats, which resets /all/ peak memory stats.

ZZono50 <@205854764540362752> What's the best way to use open source to transcribe a 2 h...

Furkan Gözükara SECourses•3/9/24, 3:02 AM

whisper

Bbgtubber Hello, guys. New member here. 🙂 I was eager to try Supir locally and found thi...

Furkan Gözükara SECourses•3/9/24, 3:02 AM

hey welcome

Furkan Gözükara SECourses•3/9/24, 3:02 AM

please reinstall and dont make parent folder name SUPIR

Furkan Gözükara SECourses•3/9/24, 3:02 AM

when both folder names are supir and SUPIR python gives error

Stage 1 now works fine. But in stage 2 I have this error: "Input type (torch.cuda.FloatTensor) and w

Similar Threads