File size: 924 Bytes
f2532fa
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
---
title: Lecture Processor
emoji: "\U0001F393"
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: 5.15.0
app_file: app.py
pinned: false
license: mit
hardware: zero-a10g
---

# Lecture Processor

Transcribe, summarize, and generate quizzes from lecture recordings using **WhisperX** and a fine-tuned **Gemma 3 4B** model.

## How It Works

1. Paste a YouTube lecture URL
2. The pipeline automatically:
   - **Transcribes** speech using WhisperX
   - **Summarizes** the lecture with structured sections (Summary, Key Points, Action Points)
   - **Generates quiz questions** (5 MCQ + 3 short answer)

## Tech Stack

- **WhisperX** - Speech-to-text transcription
- **Gemma 3 4B Instruct** - Fine-tuned with QLoRA for lecture summarization and quiz generation
- **LoRA Adapter** - [noufwithy/gemma-lecture-adapter](https://huggingface.co/noufwithy/gemma-lecture-adapter)
- **Gradio** - Web interface with ZeroGPU support