This repo includes the tools based on OpenAI, LangChain and Streamlit to increase working efficiency.
- Auto Meeting Minutes is a tool to chat with video/audio and output the meeting minutes with build-in prompt.
- Casual Chat is a ChatGPT like ChatBot.
- FilesChat is a bot to chat with your own documents, supported format including PDF, PPTX, DOCX, TXT.
- AI translator is a chatbot to translate every words you input between Chinese and English.
- CSV Chatbot is a chatbot to analyse tabula data, still in developing...
-
Install ffmpeg, ffmpeg official website.
-
Install the required dependencies:
pip install -r requirements.txt
- Run the Streamlit application:
streamlit run .\home_page.py
To configure the application, you will need to create a key.txt
file containing your Azure OpenAI API key and a config.json
file with your desired settings.
Create a file named key.txt
and add your Azure OpenAI API key, BING_SUBSCRIPTION_KEY(optional) and AZURE_COGNITIVE_SEARCH_API_KEY(optional).
Create a config.json
file with the following configuration:
{
"CHATGPT_MODEL": "xxxxx",
"OPENAI_API_BASE": "https://xxxxxx.openai.azure.com/",
"OPENAI_API_VERSION": "xxxxx",
"EMBEDDING_MODEL": "xxxxx",
"EMBEDDING_MODEL_VERSION": "xxxxx",
"BING_SEARCH_URL": "https://api.bing.microsoft.com/v7.0/search",
"AZURE_COGNITIVE_SEARCH_SERVICE_NAME": "xxxxxxxxxxxx",
"AZURE_COGNITIVE_SEARCH_INDEX_NAME": "xxxxxxxxxx"
}
- Ffmpeg is used to convert video to audio;
- Faster-Whisper is used to convert audio to subtitle;
- AgglomerativeClustering from sklearn is used to distinguish the speakers;
- OpenAI is used to do the meeting minute summary.
- Azure OpenAI GPT3.5 as the server;
- LangChain is used to create the conversational Chatbot;
- Streamlit is used to create the web-ui
- Azure OpenAI as server and LangChain used to create a QA chatbot over documents.
- Azure OpenAI as server and LangChain used to create a chatbot to create python code based on user's query, and give output.