写了个 B 站视频转音频的小工具

平时在 B 站听翻唱、长音频之类的，想存下来当音频听，试了几个工具都不太顺手，就自己用 Node 写了一个 CLI

干嘛的

把 B 站视频的音频轨提出来，转成 mp3 / flac 这些格式。支持单个视频和合集批量下载。

bili-audio-downloader https://www.bilibili.com/video/BVxxxx

为什么不用 yt-dlp

试过，但要装 Python 和一堆依赖，参数也多，B 站的 cookie、音质这些还得自己折腾。我就想要一个 npm install -g 完事直接用的东西。

大概流程

请求视频页面 → 从 HTML 里扒 window.__playinfo__ → 找音频流 → 选最高码率 → 下载 → ffmpeg 转格式。就这么几步。在科研的时候也考虑过调官方API，但还是选了我认为比较简单的办法，扒页面分析

一些细节

音频流选择：B 站返回多条音频流带码率，直接排序取最大的。

URL fallback：主 URL 偶尔会挂，所以 baseUrl 和 backupUrl 都收集起来挨个试。

格式转换：先下成 m4a，再用 ffmpeg 转目标格式为mp3，不转的话虽然是m4s但还是视频格式，老设备读不出。

Cookie：需要高码率音频，做了手动输入 + 本地保存 + 下次自动加载，不优雅但能用。

合集下载：通过 /x/polymer/web-space/seasons_archives_list 分页拿数据，做了个简单的预览菜单，看完再决定下不下。

并发：用 p-limit 控制（默认 3 个），配 cli-progress 做多进度条，不会太慢也不会被限流。

用到的东西

Node.js / TypeScript、axios、ffmpeg-static、cli-progress、p-limit、readline，没用什么重框架。

安装

npm install -g bili-audio-downloader

已知问题

依赖 B 站页面结构，改了就得跟着修
cookie 要自己弄
ffmpeg 算是个外部依赖

后续可能做的

自动读浏览器 cookie
补 metadata（封面、标题）
简单 GUI（可能用 Tauri）

项目不复杂，但 B 站接口处理、音频流 fallback、CLI 交互这些细节还是花了点时间。有类似需求的可以试试。

I Built a CLI Tool to Download Audio from Bilibili

I often listen to covers and long-form audio on Bilibili and wanted to save them as audio files locally. After trying a few existing tools that didn’t quite fit my workflow, I wrote my own CLI tool in Node.js.

What It Does

Extracts the audio track from a Bilibili video and converts it to formats like mp3 or flac. Supports both single videos and batch downloading from playlists.

bili-audio-downloader https://www.bilibili.com/video/BVxxxx

Why Not Just Use yt-dlp?

I tried it, but it requires Python and a bunch of dependencies, has a lot of flags to configure, and getting Bilibili cookies and audio quality right still takes manual effort. I wanted something where npm install -g is all you need.

How It Works

Request the video page → scrape window.__playinfo__ from the HTML → find the audio stream → pick the highest bitrate → download → convert with ffmpeg. That’s basically it. I considered using the official API but went with page scraping as the simpler path.

Implementation Details

Audio stream selection: Bilibili returns multiple audio streams with different bitrates. I sort them and take the highest.

URL fallback: The primary URL occasionally fails, so both baseUrl and backupUrl are collected and tried in sequence.

Format conversion: The raw download is an m4a file, which gets converted to the target format (e.g. mp3) via ffmpeg. Without conversion, older devices can’t read it even though the container is technically audio.

Cookie handling: High-bitrate audio requires authentication. I implemented manual cookie input with local storage and auto-loading on subsequent runs — not elegant, but it works.

Playlist downloading: Uses the /x/polymer/web-space/seasons_archives_list endpoint with pagination. A preview menu is shown before downloading so you can decide what to grab.

Concurrency: Controlled with p-limit (default: 3 concurrent downloads), paired with cli-progress for multi-bar progress display. Fast enough without hitting rate limits.

Stack

Node.js / TypeScript, axios, ffmpeg-static, cli-progress, p-limit, readline. No heavy frameworks.

Installation

npm install -g bili-audio-downloader

Known Issues

Depends on Bilibili’s page structure — breaks if they change it
Cookie setup is manual
ffmpeg is an external dependency

Possible Future Work

Auto-read cookies from the browser
Embed metadata (cover art, title)
Simple GUI (maybe with Tauri)

The project itself isn’t complex, but the Bilibili API handling, audio stream fallback logic, and CLI interaction took some time to get right. Give it a try if you have similar needs.

b站视频转音频

Bilibili Video to Audio Downloader