muziqa 1.0.23

App for various music-file-collection analyses (for example, chart of most popular artists in your MP3 collection) and AI playlist generation

muziqa

Analyze your music collection and generate interesting charts using command-line:

• Top artists by track count
• Tracks by decade
• Tracks by year, with a 5-year rolling average of mean tracks per artist
• Tracks by country (optional, see below)
• Tracks by genre (optional, see below)

Also create MP3 playlist using Anthropic Claude LLM. The generated playlist is an MP3 file of your songs, picked by AI, per your prompt. Example:

$ muziqa Music --playlist "Playlist of rock tunes where lead guitar is as close as possible to the picking style of Mark Knopfler. Max 1 hour. Max one song per artist." --playlist-output mark.mp3

Install

Linux / Mac:

$ pipx install muziqa

Windows (PowerShell; NOTE: untested):

winget install Python.Python.3   # if Python not already installed
pip install pipx
pipx ensurepath                  # restart terminal after this
pipx install muziqa

Windows + --playlist: also install ffmpeg: winget install ffmpeg

Usage

No API keys, no accounts, no streaming service — just point muziqa to a folder of music files:

$ muziqa /path/to/music

Reads tags from all supported files in the folder and subfolders, and saves two charts:

• muziqa.png — top artists + tracks by decade
• muziqa_years.png — tracks by year with rolling average

Supported formats: MP3, FLAC, WAV, M4A, OGG

Country and genre charts

$ muziqa /path/to/music --country
$ muziqa /path/to/music --genre
$ muziqa /path/to/music --country --genre

Looks up each artist's country of origin and genre from MusicBrainz and saves additional charts:

• muziqa_country.png — tracks by country
• muziqa_genre.png — tracks by genre

Note: The first run with --country or --genre queries MusicBrainz for every unique artist at 1 request/second (required by their API). For a large collection this can take a bit of time. Using both flags together does not double the time — data is fetched in a single pass. Results are cached in muziqa_mb_cache.json so subsequent runs are instant.

Note on "Unknown" genre: This means MusicBrainz either didn't find the artist or has no community-submitted genre tags for them. It does not affect the rest of the chart.

All options

Option	Description
DIR	Directory of music files to analyze
--flat	Search only the given folder, not subfolders
--country	Fetch artist countries from MusicBrainz and plot by country
--genre	Fetch artist genres from MusicBrainz and plot by genre
--output FILE	Output image filename (default: muziqa.png)
--top N	Number of top entries to show (default: 20)
-playlist DESC	Create a playlist MP3 matching the given description (requires ANTHROPIC_API_KEY and ffmpeg)
--playlist-output FILE	Output file for playlist (default: playlist.mp3)
--model MODEL	Claude model for --playlist (default: claude-sonnet-4-6). Tip: use 'llm-models -p Anthropic' to list available models --> github.com/ljbuturovic/llm-models

Examples

$ muziqa ~/Music
$ muziqa ~/Music --flat
$ muziqa ~/Music --country --genre
$ muziqa ~/Music --top 30 --output top30.png