I’m doing the same thing, it’s revolutionary.
But I still suck at it. When I get good I will share my prompt.
Yeah Ive wasted days of my life trying to edit the Bandcamp file formats (which I hate - as they always stick the album name twice in the track name)
But yes, do share a master prompt if you get there.
I’m doing it section by section as still paranoid I’m doing to do something to completely f*k it all up ! But so far, so good 
how does Claude change file names hosted on your computer? I’d love to know how to do everything you’ve implemented there.
If you check out the Claude threads on reddit, there are people with vast knowledge explaining why Claude is now very quick to end a session compared with a month or 2 ago, even at the so called “pro” tier. It might be the model we see all AI moving forward with - dangle the carrot and get people onboard for $20 a month, then make that insufferable and force the to pay for the $100 per month tier to properly access he features.
EDIT: I just asked Claude how it would achieve the file formatting and it gave me the following:
The two realistic approaches
The cleanest path is Claude Code in terminal, where Claude writes and executes the scripts against your actual files in real time. The alternative is Claude generating Python scripts you run yourself. Either way, the underlying tools are the same.
Per task breakdown
ID3 tags (and equivalent for AIFF/FLAC): The Python library mutagen handles read/write across all major formats. Claude can write a script that reads existing tags, cross-references MusicBrainz or Discogs via their APIs for correct metadata (artist, album, year, genre, label), and writes it back. beets is a fully-formed CLI tool built on this idea, and Claude can help configure and extend it.
File renaming: Pure Python string manipulation once you define the rules. Strip underscores, remove leading track numbers (01_, 01 - ), normalise capitalisation, handle featuring credits. Claude writes a dry-run version first so you see proposed renames before anything changes.
Artwork: mutagen handles embedding. The source can be a local folder of artwork files matched by album name, or fetched from MusicBrainz Cover Art Archive or Last.fm API. Claude can write a script that checks whether artwork is already embedded before doing anything.
Format conversion (WAV/FLAC to AIFF): ffmpeg is the tool, and Claude can wrap it in a Python script that batch-processes a folder, preserves tags (FLAC tags survive; WAV is trickier since WAV tag support varies), and optionally deletes the source file after confirming a clean conversion. One command structure:
ffmpeg -i input.flac -c:a pcm_s24be -f aiff output.aiff
Claude would generate the full batch wrapper around this.
Finding duplicates: Two approaches depending on how thorough you want to be. Hash comparison (MD5/SHA) catches exact byte-for-byte duplicates. Audio fingerprinting via chromaprint/pyacoustid catches the same recording in different formats or at different bitrates, which is usually what you actually want. Claude can script either, outputting a report before any deletion happens.
Practical starting point
The most useful first move is probably a Claude Code session where you point it at a sample folder of 20-30 files and say “audit these: check tag completeness, flag naming inconsistencies, identify any duplicates, list what’s missing artwork.” That gives you a picture of the problem before writing anything destructive.
If you’re not running Claude Code yet, the alternative is asking Claude here to generate a standalone Python script for any one of these tasks. The format conversion and duplicate finder are probably the highest-value starting points given how tedious both are manually.
1 Like
Claude Cowork is essential no make this work
my main gripe of late has been that when the car audio plays a USB stick it defaults to alphabetical track order! Am sure there is a way to automatically fix this via settings but have yet to find it.
on the plus side, did manage to retrieve ancient files from an ipod using the sharepod app (on windows 11) although the free version limits you to small batches with 25 second countdown delays in between. Simply plugging the ipod into newer cars doesn’t seem to work anymore alas.
Yes its definitely Claude Cowork that you need.
You can then connect Claude Cowork to files on your hard drive (or USB drive), which allows it to change or make new files.
Then when you prompt it (eg to change file names or change .wav to .aiff etc), it figures out what applications it needs (e.g. Python) and does it all automatically. So no need to code etc.
I’m on the $20 a month ‘pro’ account and you do hit the limits quite quickly.
But is a bit of a game changer
I posted a couple of slightly off topic links a long way above, but they might help you.
Advanced Renamer is a free tool for batch renaming of multiple files, particularly good for swapping out characters, renaming files based on ID tag fields, swapping fields around etc.
Tagscanner can do the same (although not so good for batch) but its main purpose is editing the ID tags. like removing someone elses annoying ‘genre’ tag. It does a lot of advanced stuff I don’t use. Also free but PC only.
I use both all the time
Foobar2000 is the best application I found for bulk filename formatting, conversion and consistent tagging etc Takes a bit of setup and messing with string formatting rules, and the app is pretty outdated but very powerful eg. has plugins for BPM detection. But obviously AI is very helpful here and less rigid with input. You might be able to use a combination of the 2, to get around limits.
1 Like
Be careful doing mass file actions with any form of AI. Even the experts can get it wrong. We see this in all sorts of AI interactions where it runs out of resources for bigger actions and just gives up or lies. What works for a test sample of 20 may not work on a bigger sample
This all sounds like witchcraft to me, I never knew such tools existed. Would be useful to ID duplicate files though, which one is the best for that?
1 Like
Yeah I’ve been doing this in batches in Claude Cowork as not brave enough to do to entire database in case something happens. And I’ve made sure I had a backup first !
But Claude has been great at sense checking things and spotting anomalies.
And I don’t think it can actually delete any files. So if for instance you ask it to scan for non .AIFF files and convert > it will create a new folder e.g. WAV_to_be_deleted of all the original files. So you can check and manually delete.
Just finding it a lot more intuitive than a tool
I’ve been using the free mac software XLD since I’ve been dealing with digital files ever and it’s really easy to use with a visual interface. You are basically one click away to convert all your folder to the desired format file.
In my case I have a folder of incoming files that I gather from different places (rips, promos, downloads, etc - that are flac, wav, other bit formats etc.) then I have a output folder where once i click convert (always to aiff) they just land there.
From then on I do other things but the converting parts is really easy and no AI needed for it. Can only recommend. If someone is interested I can send some screenshots with the exact configs.
For renaming and metadata I guess it can be helpful though!
2 Likes
For me it’s mainly getting rid of duplicates and crap mp3 files that say 320kbps but are less (Claude struggles with this bit, for now). Later I want to fix tags and playlists. I’m interested to see how it’ll play with Rekordbox…
2 Likes
What is the Claude prompt to make this possible?
same, I also have not yet figured out a way to avoid identical files from Bandcamp Pre orders/full release. anyone got tips?
My audio library is duplicated in 3 places: Google Drive, 12TB SATA Ironwolf Pro, 4TB Samsung 990 Pro
I would still love a way to access my iTunes library from my Mac mini to my iPhone. Surely there is a way to do this without subscribing to Apple Music?
I asked this question further up the thread. Not found an ideal solution but some of the responses might be useful
That’s iTunes Match
I thinks it’s about £25/year so it is subscription but only a small one. I’ve used it for ages
It just works, but its matching algorithm can play havoc with artwork and naming. Particularly for compilations which it matches with Apple Music at the point you upload but are later pulled from the platform. At this point it tends to match your tracks to other versions of the same song eg from the original artist album rather than your comp.
200mb (actually about 210mb) file size limit so I have had to split numerous mixes into parts or reduce to 256 bitrate to get them to fit.
But for all that it gives me all my mixes and tracks that don’t appear in Apple Music right on my phone
1 Like