-
Notifications
You must be signed in to change notification settings - Fork 0
Ama-Annor/Interview-Insights
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Repository files navigation
LINK TO APP DEMO: https://youtu.be/PYmgFxYyTkQ # Interview Insight Generator This Colab notebook contains a Streamlit app that transcribes audio and video interviews, generates insights, and answers follow-up questions using AI. ## Features 1. Audio/Video Transcription: Converts uploaded audio or video files into text using OpenAI's Whisper model. 2. Insight Generation: Analyzes the transcription to provide key insights using a BART language model. 3. Follow-up Questions: Allows users to ask questions about the transcribed content and receive AI-generated answers. ## Setup 1. Run the first cell to install required packages: - openai-whisper - moviepy - streamlit - pyngrok - transformers - torch 2. The second cell contains the main application code (`app.py`). 3. Set up ngrok for tunneling: - Run the cell to set the ngrok authtoken - Set the NGROK_AUTHTOKEN environment variable 4. Launch the Streamlit app and create an ngrok tunnel to make it accessible. ## Usage 1. Upload an audio or video file (supported formats: wav, mp3, m4a, mp4, avi, mov, mkv). 2. The app will transcribe the file and display the transcription. 3. Choose insight generation options: - Insight type (general, technical, business) - Number of insights - Content type (e.g., interview, presentation) 4. Generate insights based on the transcription. 5. Ask follow-up questions about the content. ## Components - `convert_to_audio()`: Converts video files to audio if necessary. - `transcribe_audio()`: Uses Whisper model to transcribe audio to text. - `generate_response()`: Generates responses using the BART model. - `generate_insights()`: Extracts and formats insights from the transcription. - `answer_followup()`: Generates answers to follow-up questions. ## Deployment The notebook uses ngrok to create a public URL for the Streamlit app running in Colab. This allows for temporary deployment and sharing of the app. ## Notes - This setup is suitable for testing and demonstrations, not for permanent deployment. - The ngrok free tier has limitations on simultaneous sessions. - Colab has usage limits and may disconnect after periods of inactivity. ## Creators Created by Ama Annor and Claude Quartey
About
Using an LLM to build an app that will generate high quality insights from an interview. The app would give deep insights on interviews.
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published