zuchenglong/vibe-notes
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
A curated corpus of 4759 Douyin video transcripts with AI-generated summaries
3 commits 1 branch 0 tags 20 MiB
  • Python 100%
Find a file
Exact
zuchenglong 555df1540d proofread: Pass 2 batch rewrite of 4,288 transcripts 
- Pass 1 (逐字稿改写): adds punctuation, fixes homophones, splits paragraphs
- Pass 2 (AI 概要重做): rewrites summaries to focus on information increment
- Tools added under tools/proofread/ for reproducibility

Coverage:
- Pass 1: 4,551 / 4,759 files (95.6%)
- Pass 2: 3,582 / 4,759 files (75.3%; remaining 25% had no original summary)

Per-file QA diffs are kept locally (*.diff.md, *.summary.diff.md) and
gitignored.
2026-07-15 22:00:19 +08:00
+1的销售人生 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
AI proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
tools/proofread proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
一只耍废的猫 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
周期的巨轮 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
商业 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
小司频道 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
悦润禾 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
执戟郎 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
无限进化的袁Sir proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
罗永浩-播客正片合集 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
职场 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
西川的书房 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
认知 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
讲政堂 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
逻辑模组Eva proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
金融 proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
.gitattributes init: 4759 Douyin video transcripts with AI summaries 2026-07-13 22:32:58 +08:00
.gitignore proofread: Pass 2 batch rewrite of 4,288 transcripts 2026-07-15 22:00:19 +08:00
index.csv init: 4759 Douyin video transcripts with AI summaries 2026-07-13 22:32:58 +08:00
README.md init: 4759 Douyin video transcripts with AI summaries 2026-07-13 22:32:58 +08:00

README.md

vibe-notes

A curated corpus of Douyin (TikTok China) video transcripts, organized by source.

Overview

This repository contains 4,759 transcribed videos from 16 Douyin creators, with AI-generated summaries covering key topics, viewpoints, and golden quotes.

Stats

  • Total files: 4,759 Markdown documents
  • Total characters: ~6.7M (Chinese)
  • AI summaries: 3,964 / 4,759 (83%)
  • Update: 2026-07

Directory structure

Directory Files Topics
+1的销售人生 1172 AI: 377
AI 364 AI: 364
一只耍废的猫 198 AI: 198
周期的巨轮 79 AI: 79
商业 51 AI: 51
小司频道 2222 AI: 2222
悦润禾 26 AI: 26
执戟郎 193 AI: 193
无限进化的袁Sir 50 AI: 50
罗永浩-播客正片合集 30 AI: 30
职场 15 AI: 15
西川的书房 23 AI: 23
认知 138 AI: 138
讲政堂 120 AI: 120
逻辑模组Eva 42 AI: 42
金融 36 AI: 36

File format

Each .md follows this structure:

# <video title>

**话题标签**: `#tag1` `#tag2`
**视频时长**: M:SS
**抖音ID**: 1234567890
**源文件**: `<filename>.mp3`

## AI 概要

**核心主题**: ...
**核心观点**: ...
**金句**: ...
**关键词**: ...

---

## 逐字稿

<full transcript>

Index

See index.csv for searchable index of all 4,759 files.

Pipeline

  1. Transcription: faster-whisper large-v3 + NVIDIA CUDA float16
  2. Punctuation: timestamp-pause + Chinese particle rules
  3. AI summary: MiniMax-Text-01 (Anthropic-compatible API), 6-thread parallel

License

Personal use. Source content belongs to the original Douyin creators.