Report Bug · Request Feature
- Add support for Whisper API
- Translator
The goal of this project was to make learning new things easier. The program was created for my own use, but I provide it with full support in the future.
This project allows you to create notes or summarize using LOCAL artificial intelligence models. The program does not require a penny to run (unless we are talking about a slightly better computer)
The user can enter text manually, provide a link to an article, a YouTube video or a file in .PDF or .TXT format
Then, after initial data processing, the program allows you to create notes/summaries using artificial intelligence.
As I mentioned, the project allows you to provide various data sources. I will try to briefly discuss each of them.
Here, the user can manually enter the text that he would like to process
Thanks to the Selenium framework and page content parsing, it is enough to provide a link to any article on the Internet and the program will automatically download the entire page content and then save all the text contained in the h1, h2, h3, h4, h5 and p
In this case, two solutions were used. First, the program checks whether subtitles have already been added manually or generated for this film. Most videos have subtitles turned on, but there are some without them. Then the program automatically downloads the movie as an .mp3 file and then divides it into parts - 2.5 minutes each. Then, using the Whisper model, which also works locally and does not require any costs, it creates a transcription of the entire video.
Here the principle of operation is similar to Text input
, the file is saved and then it is parsed to extract all the text from it
Everything is done by using artificial intelligence models that run locally on the computer.
- Python
- FastAPI
- Selenium
- Beautiful Soup 4
- SQLite
- Docker
- Javascript
- HTML, CSS, Bootstrap
- Powershell
- Ollama
- Whisper
- OpenAI
- Groq
git clone https://github.com/DEENUU1/property-aggregator.git
cp .env_example .env
LLM_MODEL
(OPTIONAL) If you want to use Ollama here you can select one of model from this list https://ollama.com/library (codegemma, gemma, llama2, mistral, ...)
OPENAI_APIKEY
(OPTIONAL) If you want to use OpenAI with models (GPT4/GPT3.5-Turbo) here you can add your API KEY
GROQ_APIKEY
(OPTIONAL) If you want to use GROQ API with models (Llama3/Llama2 etc.) here you can add API KEY
WHISPER_MODEL
(OPTIONAL) here you can select one of model from this list https://github.com/openai/whisper (tiny, base, small, medium, large)
It's available on Windows, MacOS and Linux https://ffmpeg.org/download.html
# Build Docker image
docker build -t app .
# Run container
docker run app
This is a great option if you are using Windows.
Right click on run.ps1
file and click Run With Powershell
pip install -r requirements.txt
cd ./src
python app.py
See LICENSE.txt
for more information.