Turn a folder of unknown .aif files into tagged MP3s that look album-ready again.
Feta is a command-line workflow for the awkward moment after the rip. You point it at a collection folder, narrow the intake by filename prefix, fingerprint each file through AcoustID, resolve artist and album context through MusicBrainz, preview cover art in the terminal, then let ffmpeg write clean MP3s with metadata and optional artwork.
One command, four useful decisions, and a fallback when the metadata gets fuzzy.
The current repository is not a vague “audio AI” idea. It has a clear operational arc: collect files from a folder, match what can be identified, ask the user to confirm the consensus artist and album, then finish the batch with artwork and a rescue path for unmatched items.
Start with the actual pile, not a fantasy importer.
Feta remembers previous collection paths in state.json, lets you enter a new folder when needed, and only scans files that match the selected extension and prefix.
- Loads prior collection paths and reuses them if you want less typing.
- Offers a single current intake type:
.aif. - Applies a case-insensitive filename prefix filter before any API lookups.
The landing page can be fun, but the product posture should stay practical: you are usually cleaning up a real folder of half-labeled files, not exploring a media library for sport.
What a credible pass looks like from prompt to finished folder.
The copy below mirrors the repo’s current behavior rather than an imagined future release: the convert flow is collection-based, artist and album are confirmed after metadata lookup, and cover art can be accepted, rejected, or replaced with a custom image URL.
Collection folder path is either reused from prior runs or entered fresh, then saved back to state.json.
Input file type currently offers one deliberate lane: .aif.
What do the filenames start with? lets you isolate a side, take, or subset inside a larger dump without moving files around first.
The run asks you to confirm the most common matched artist and album, then asks again about artwork before it writes the final folder.
The whole point is to make the unknown pile legible again.
This page positions Feta as a sharp, characterful CLI for audio clean-up rather than a generic audio utility. The actual differentiator is not spectacle. It is the calm sequence: filter the folder, fingerprint the files, confirm the common metadata, preview the art, and ship the usable batch.
Build note from the repo review: the current code exposes a convert command and sets the Commander program name to Feta. The landing page keeps the public story focused on Feta while staying honest about the present command surface.