AI 仓库情报摘要

FR-AI / ANALYSIS

为什么值得关注

它整合了本地 AI（Ollama）、免费 TTS（Edge-TTS）和免费素材库（Pexels）到一条自动化流水线中，无需付费 API 或云服务。

适合谁使用

需要制作讲座视频的教育工作者和培训师
需要快速生成带旁白和视频的内容创作者
希望探索 AI 生成媒体的学生或爱好者
寻找可定制的开源视频生成工具的开发者

典型使用场景

根据单个主题提示生成教育课程视频
为社交媒体制作带素材的旁白幻灯片
利用不同 TTS 语音快速创建多语种讲座原型
自动化在线学习平台的视频制作流程

项目优势

完全本地运行的大模型和 TTS – 无周期性 API 费用，数据隐私有保障
双界面（Web UI + 命令行）支持实时进度跟踪和预览
从脚本生成到最终 MP4 的自动化流水线，包含字幕叠加
素材缺失时自动回退为纯色背景，且渲染前可编辑脚本

使用前须知

需要本地安装 Ollama、FFmpeg 并获取 Pexels API 密钥
脚本质量取决于所选 Ollama 模型的大小和能力
素材仅来自 Pexels 提供的资源，不支持自定义图片/视频上传

README 快速开始

🎓 Lecture Video Maker

An AI-powered tool that converts any topic into a full lecture video with voiceover, stock visuals, and subtitles — all from a beautiful web UI or the terminal.

✨ Features

AI Script Generation — Uses your local Ollama model to write a multi-segment lecture script
Natural Voiceover — Microsoft Edge-TTS synthesizes speech (no API costs)
Stock Visuals — Automatically fetches relevant videos and images from Pexels
Subtitle Bar — Pillow renders animated subtitle overlays (no libass needed)
Web UI — Full browser-based interface with live progress tracking and video preview
CLI Mode — Fully interactive terminal alternative

📋 Requirements

System

Python 3.8+
FFmpeg (must be in PATH)
Ollama running locally with at least one model pulled

Python Packages

pip install edge-tts requests pillow numpy

🚀 Quick Start

1. Clone the repo

git clone https://github.com/YOUR_USERNAME/lecture-video-maker.git
cd lecture-video-maker

2. Install Python dependencies

pip install edge-tts requests pillow numpy

3. Start Ollama and pull a model

ollama serve          # in a separate terminal
ollama pull llama3    # or mistral, gemma2, etc.

4. Get a free Pexels API key

5. Launch the Web UI

python lecture_gen.py

Opens automatically at http://localhost:8080

Or use CLI mode

python lecture_gen.py --cli

🖥️ Web UI Walkthrough

Settings Card — Enter your Pexels API key, Ollama URL, and select a voice
Topic & Parameters — Set your lecture topic, choose duration and media preference
Edit Script — Review/edit the AI-generated narration per segment before rendering
Render — Watch live progress with log output
Download — Preview and download the final .mp4

📁 Output Structure

lecture_output/
├── _.mp4       ← final lecture video
└── /
    ├── script.txt                ← full generated script
    ├── audio/seg_00.mp3 …        ← TTS audio per segment
    ├── media/visual_00.mp4 …     ← downloaded Pexels media
    ├── subs/seg_00/frame_*.png   ← subtitle frames
    └── segments/se

项目描述

AI-powered lecture video generator: Ollama + Edge-TTS + Pexels + FFmpeg. Web UI with live progress tracking.

lecture-video-maker