Bilibili YouTube Video Summarizer

Problem

Engineers often need more than a saved link when they encounter a Bilibili or YouTube video. They want the actual claims, data points, and a usable Chinese version when the source is not in Chinese. Manually downloading audio, running transcription, and organizing text can get messy, especially when Bilibili short links and YouTube subtitle routes require different handling.

How It Works

bilibili-youtube-video-summarizer follows the path “video → audio/subtitles → transcript → translation → structured summary.” Key capabilities include:
- Detecting Bilibili and YouTube URLs and choosing between audio download and subtitle retrieval
- Preferring the YouTube subtitle API when available, then falling back to audio download
- Transcribing with faster-whisper by default, supporting txt, md, srt, and vtt, with openai-whisper as an optional fallback
- Translating non-Chinese content into fluent Simplified Chinese and producing a key-term comparison

The workflow starts with an environment check for Python >= 3.10, ffmpeg, faster-whisper, plus platform-specific dependencies. It then parses the URL, selects the content route, generates or reads transcripts, and produces a summary with basic info, core points, key data, and actionable suggestions where applicable.

Limits To Note

This is best for converting a single video into a searchable, readable document rather than bulk collection. Bilibili short links may hit anti-bot behavior and should be resolved to real BV URLs first. The YouTube no-subtitle path may require a logged-in browser session. For Chinese videos, explicitly setting --language zh improves accuracy, and low-quality audio is usually enough for transcription.