WhisperAI (local)

4 месеца ago
tasko1
1 minute

Web site: https://whisperai.com/

Git Hub: https://github.com/openai/whisper

Test and work envirement:

First working test environment

OS : Windows 11 Pro Build 26220 10.0.26220
GPU: RTX 4060m
CPU: i5-13500H

A) On Widnows Cmd (win+R => cmd)

01. winget install -e –id Git.Git
02. winget install -e –id Anaconda.Miniconda3 –version py312_25.1.1-2 –override „/AddToPath=1“
03. winget install -e –id Gyan.FFmpeg –version 7.0.2

B) Change conda env folder
– copy file .condarc to C:\Users\<user name>\ or create in manual with folling structure

envs_dirs:
– E:\Programs\Conda\env

pkgs_dirs:
– E:\Programs\Conda\cache

On Anaconda Promt (under admin rights)

01. conda create –name Whisper-WebUI python=3.10
02. conda activate facefusion
03. conda install nvidia/label/cuda-12.9.1::cuda-runtime nvidia/label/cudnn-9.10.0::cudnn
– akternative by old gpu: conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia
04. E:/
05. cd Programs
06. git clone https://github.com/jhj0517/Whisper-WebUI.git
07. cd Whisper-WebUI
08. pip install -r requirements.txt
09. python app.py

Can be make bat files for easy start:

cd E:\Programs\Whisper-WebUI
venv\Scripts\activate
python app.py

Or just start over Anaconda Navigator