Audio Organization

hey MPU,

Need some help here, been bouncing ideas off of ChatGPT, scribbled away endlessly at notebooks, but still scratching my head. There’s been an update since this old thread

We are nearing completion of the app, the devs have provided for me to insert keywords and tags to the files. All that has been resolved.

My current issue is ‘Albums’ vs ‘Playlists’.
My earlier definitions have historically been Albums are ‘closed’ (you don’t really add more to it) and Playlists are ‘open’ (they get changed whenever you feel the need to create or add)

For most of the speakers (artists) in the library, I can figure out, and give them album names (except in these cases they will be ‘open’ and really closed)

However, I have a handful of speakers that have 2,000+ audio talks they have given. There is one in particular in my collection who gives a talk in a different location/county every day (his diocese is very large and covers multiple states). As of right now, I treat each of his talks (unless they are a series) I treat them all as ‘singles’…however from a user perspective it can get daunting to scroll through a list of 2,000 albums that contain only 1 talk.

I thought about maybe organizing them into years? 2023 Talks, 2024 Talks…unfortunately, not all of these audio files have the date stored in the title of metadata.

Anyone have suggestions or solutions? I know it’s a very unique problem, and the answer could be the user could just ‘search’ for what they want. But what if the user wants to see what else this artist has spoken about.


  • This will continue to happen as I upload regularly
  • There is at least 4-6 artists this will affect who have large audio files
  • How to make it user-friendly so it’s not a doom scroll

I could create thematic albums and hopefully try to find some connections via the title?
Any other thoughts?

Decades ago I set a final year degree computing science project called MERLIN (My Entire Record Library Indexed Neatly). Student attained a first class honours degree for it and their other work. Sadly the resultant code fell into disuse as a) it was written in COBOL-68 for Tops-10 on a DECsystem-10 and b) used the CODASYL database (Codd’s seminal paper had either not then been published or only about to see the light of day).

As my record collection is primarily classical the database structure and the reports generated mimic the now defunct annual Gramophone Record Catalogue taxonomy of musical genres.

Still have the student’s report somewhere in my personal archives but while there is an open-source Cobol compiler available there is no corresponding re-implementation of CODASYL. Supposed I could use a Raspberry Pi running a DECsystem-10 emulator to revive the code

and covered by The Guardian newspaper

I did use TS10 on my decommissioned Linux machine.

1 Like

Not really an answer to the question, but perhaps a different way to approach the problem:

For a while, Flickr (a photo sharing site for those who are not familiar) espoused the use of what they called “machine tags”.

They were very early implementers of plain tags for photos — dog, sunrise, rain — and machine tags were a simple variation on these.

Distinct from the above examples, machine tags would be like — animal:dog, time:sunrise, weather:rain — in other words, each value was assigned to a domain. This was completely freeform for the user. There was nothing stopping me from using weather:dog or even dog:cat. I’m not even convinced they changed anything on the back end

So I can imagine speaker:LBJ, year:2020, topic:rainbows. With some forethought on both the domains and any (self-imposed) restrictions on their values (like sticking to initials or full names) then a generic means of organisation would allow the user to pick any domain to group by or filter on.