Foobar2000 AI元数据处理插件

最近，用AI写了个fb2k的元数据处理插件，用来：

1. 刮削歌曲元数据（图片暂时还不支持）；
2. 翻译非中文（日语、韩语等）标题、专辑等字段为中文；
3. 识别歌曲版本号、曲风；

目前算是一个MPV版本吧，简陋但可用。至少满足了我的需要。

下载连接。（https://files.cnblogs.com/files/ztcheck/foo_ai_metadata.zip?t=1774613606&download=true）

下面是说明。

foo_ai_metadata

A foobar2000 plugin that uses AI to automatically analyze, scrape, and enhance music metadata.

Features

• Stage 1: Metadata Scraping - Automatically fetches metadata from MusicBrainz, Discogs, and AI
• Stage 2: Enhancement - AI-powered translation, genre classification, and edition identification
• Multi-source Support - MusicBrainz (authoritative) → Discogs (supplemental) → AI (fallback)
• Smart Caching - SQLite-based cache to reduce redundant API calls
• Backup & Rollback - Automatic backup before modifications with rollback support
• Multiple AI Providers - Supports OpenRouter, Zhipu AI, Google Gemini, and Ollama (local)

Requirements

• Register and get API KEY from ZHIPU AI (https://bigmodel.cn/console/overview).This is a free ai service for some models now.
• foobar2000 2.0 or later
• Windows 10/11
• Python 3.11+ (required)
• External Tags (recommended for CUE sheet support) - Download from foobar2000 components
  • Configure under Advanced → Tagging → External Tags:
    • ◉ Use only SQLite (fastest)
    • ☑ Open properties dialog after external tag creation
    • ☑ Enable art support in external tags
    • ☑ Take over all tagging (Optional, but recommended)

Installation

1. Download the latest release from Releases
2. Extract the archive
3. Copy foo_ai_metadata.dll and foo_ai_metadata folder to your foobar2000 components folder
4. Restart foobar2000

Quick Start

1. Select one or more tracks in foobar2000
2. Right-click → AI Metadata → Stage 1: Scrape Metadata
3. Review the results and select fields to write
4. Click Apply Selected

Menu Commands

Command	Description
Stage 1: Scrape Metadata	Fetch basic metadata from MusicBrainz/Discogs/AI
Stage 2: Enhance Metadata	AI-powered translation, genre classification, edition identification
Rollback to Initial	Restore tracks to original state before AI processing
Cache Statistics	View cache hit rate and database size
Clear Cache	Clear cached metadata for selected tracks or all

Configuration

Access settings via: File → Preferences → AI Metadata

General Settings

Setting	Description
Provider	AI provider (OpenRouter, Zhipu, Gemini, Ollama)
API Key	Your API key for the selected provider
Model	AI model to use
Use Env Var	Use environment variable for API key

Python Settings

Setting	Description
Python Path	Path to Python executable (auto-detected)
Auto-install Packages	Automatically install required Python packages.Only checked at the first time, otherwise unchecked.

Cache Settings

Setting	Description
Enable Cache	Enable/disable caching
Expiration (days)	Cache entry expiration time
Max Size (MB)	Maximum cache database size
Auto Cleanup	Automatically clean expired entries

Advanced Settings

Setting	Description	Default
Task Queue Batch Size	Number of tracks per batch for Stage 1	50
AI Batch Size	Number of tracks per batch for Stage 2	10
Per-Track Timeout (sec)	Timeout per track	60

Supported Tags

Stage 1 Output

Tag	Description
TITLE	Track title
ARTIST	Artist name
ALBUM	Album name
YEAR	Release year
TRACKNUMBER	Track number
DISCNUMBER	Disc number
COMPOSER	Composer
LYRICIST	Lyricist
LABEL	Record label

Stage 2 Output

Tag	Description
GENRE	AI-classified genre
EDITION	AI-identified edition
TITLE_ZH	Chinese translation of title
ALBUM_ZH	Chinese translation of album
ARTIST_ZH	Chinese translation of artist