
Subtitle-Gen 本機影音字幕+摘要|Local Subtitles & Summaries
我長期處理中英夾雜的技術型影音內容(教學影片、會議與簡報錄音),很清楚這類素材最頭痛的就是專有名詞和英文縮寫被聽錯。這個 Skill 把我實際在用的整套做法工具化:以 faster-whisper 在本機 CPU 就能轉錄(免 GPU、免雲端 API、免費),並用「轉錄前餵專有名詞提示+轉錄後批次修正同音錯字」兩道工序把準確率拉高;YouTube 連結則走「先抓現成字幕、失敗自動改抓音訊轉錄」的穩定路線,最後一併產出 SRT 字幕、逐字稿與帶時間戳的重點摘要。全程本機運算、檔案不上雲,適合在意內容隱私又要快速消化長影音的人。
I work daily with mixed Chinese-English technical audio, where jargon and acronyms get mis-heard most. This Skill packages my workflow: faster-whisper runs locally on CPU (no GPU, no cloud API)
#影片#媒體#教育
評分
需要更多評價
已售出
1
使用方式
下載
Subtitle-Gen|本機影音轉字幕+YouTube 摘要
把沒字幕的影片、音檔或 YouTube 連結,在你自己的電腦上轉成 SRT 字幕、逐字稿與重點摘要。全程本機運算 —— 檔案不上雲、免 GPU、免雲端 API、免費模型。
✨ 它能做什麼
- 🎬 影音轉字幕:mp4 / mkv / mov / avi / mp3 / m4a / wav 直接吃,不必先抽音軌
- ▶️ YouTube 連結:先試抓現成字幕,失敗自動改抓音訊轉錄
- 📝 三種產出:
.srt字幕、_transcript.txt逐字稿、_summary.md帶時間戳重點摘要 - 🎯 聽錯修正:轉錄前餵專有名詞、轉錄後批次修正同音錯字,準確率明顯提升
- 🌏 中英夾雜 OK:自動處理中英混講內容
🔒 隱私優先
全程在本機用 faster-whisper 運算,一般 CPU 就能跑。你的影片、會議錄音、敏感內容 不會上傳任何雲端。
⚙️ 怎麼運作
- 給它一個影音檔或 YouTube 連結
- (建議)先提供領域術語清單,降低聽錯
- 本機 Whisper 轉錄 → 自動列出疑似聽錯供你確認 → 批次修正
- 一次拿到字幕 + 逐字稿 + 摘要
🚀 適合誰
- 要快速消化長影片、Podcast、會議錄音的人
- 需要把影音轉成可搜尋文字的創作者與研究者
- 在意內容隱私、不想把檔案丟雲端的人
📦 需求
- Python 3.10+、
faster-whisper(首次自動下載模型) - YouTube 功能需
yt-dlp
Subtitle-Gen — Local Subtitles & Summaries
Turn any un-subtitled video, audio file, or YouTube link into SRT subtitles, a transcript, and a key-point summary — right on your own machine. Fully local: files never leave your computer, no GPU, no cloud API, free model.
✨ What it does
- 🎬 Media → subtitles: eats mp4 / mkv / mov / avi / mp3 / m4a / wav directly — no need to extract audio first
- ▶️ YouTube links: tries existing captions first, falls back to audio transcription automatically
- 📝 Three outputs:
.srtsubtitles,_transcript.txt, and a timestamped_summary.md - 🎯 Mis-hear fixes: term prompts before transcription + batch fixes after, for a clear accuracy boost
- 🌏 Mixed-language friendly: handles mixed Chinese-English speech
🔒 Privacy first
Everything runs locally with faster-whisper on a normal CPU. Your videos, meeting recordings, and sensitive content never get uploaded to any cloud.
⚙️ How it works
- Give it a media file or a YouTube link
- (Recommended) provide a quick glossary to reduce mis-hearing
- Local Whisper transcribes → flags likely mis-hears for you to confirm → batch-fixes them
- Get subtitles + transcript + summary in one pass
🚀 Who it's for
- Anyone digesting long videos, podcasts, or meeting recordings fast
- Creators and researchers who need searchable text from audio/video
- Privacy-conscious users who don't want to upload files to the cloud
📦 Requirements
- Python 3.10+,
faster-whisper(model auto-downloads on first run) yt-dlpfor the YouTube feature


