Excel Tag DB Guide

Milonga DJ Pro — building song tags for smarter similar-song & tanda suggestions

Tagging tracks with "mood" and other attributes lets the Milonga DJ Pro recommendation engine consider the emotional arc when suggesting tandas. The most efficient way to manage tags is in a single Excel file.
This guide does not ask you to write Excel from scratch — instead you grow it gradually from a table the app exports for you, so the slow work of building a DB stays efficient and safe.

0. Core principle: Excel is the "master"

On import, a cell with a value → overwrites (Excel wins)
A blank Excel cell → keeps the existing in-app tag (manual work is never wiped by a blank)
So if you keep and update the Excel file carefully, it always remains the source of truth

The thing you invest time in is the Excel file. The app side is a "copy". Save the Excel file each time and keep backups.

1. The overall flow (round-trip workflow)

STEP 1 — Export the seed
Click ⬆ Export tags in the top bar → tango-library-tags_YYYYMMDD.csv downloads. It is "a list of every track in your library (title, orchestra, year, existing tags)", so you never type from scratch.

STEP 2 — Edit in Excel
Open that CSV in Excel → fill the 3 required tags (below) → save as Excel (.xlsx) (any filename, e.g. tango-library-tags.xlsx).

STEP 3 — Import
Click 🎭 Import tags in the top bar → choose the edited .xlsx.

STEP 4 — Check the sync report
The "Tag sync report" shown after import lists synced count / unmatched Excel rows / still-untagged tracks.
Fix unmatched rows in Excel and re-import; treat untagged tracks as the next batch.

STEP 5 — Iterate in small steps
Don't do everything at once. Start with the orchestras/tandas you use most. Repeat STEP 1–4 to grow it.

2. Columns & how to fill them

Row 1 is the header (English). Rows 2+ are one track each. Do not change the header spelling.

Column	Required	Content / how to fill
`track_id`	—	In-app ID. If present in the exported CSV, don't touch it (matched first)
`title`	◎	Track title (e.g. Bahía Blanca)
`orchestra`	◎	Orchestra (e.g. Carlos Di Sarli)
`genre`	○	`tango` / `vals` / `milonga` / `cortina`
`year`	△	Recording year (1900–2100). Applied if present
`singer`	△	Singer (blank = instrumental). Applied if present
`bpm`	△	Tempo (number). If blank, the auto-analyzed value is used
`key`	△	Key (e.g. Am, D minor, F# minor)
`primary_mood`	◎	Primary mood (one of the 13 below). Rows with this blank are not synced
`secondary_moods`	△	Secondary moods. Comma-separated, multiple allowed (e.g. `nostalgia,longing`)
`flow_type`	◎	`Descent` / `Ascent` / `Sustain` / `Bounce`
`intensity`	◎	1–5 (1 = ambient, 5 = climax)
`notes`	△	Free memo (use case, "for the opening", etc.)

Fill only the 3 required ones first: primary_mood / flow_type / intensity. title and orchestra are already filled by the export. Tracks of the same orchestra/era tend to share tags, so copy-paste speeds it up.

3. Choosing primary_mood (13 moods)

Mood (JA)	Meaning (EN)	Family	Typical use
哀愁	Aishū — wistful sorrow	melancholic	Early milonga / closing masterpieces
郷愁	Kyōshū — nostalgia	melancholic	Nostalgic / a moderate opening
諦念	Teinen — resignation	melancholic	Sense of acceptance
切望	Setsubō — yearning	longing	Craving to move forward
回想	Kaisō — reminiscence	calm	Quiet memories
安らぎ	Yasuragi — serenity	calm	Calm, settled
希望	Kibō — hope	positive	The start of a rise
情熱	Jōnetsu — passion	passionate	The core of the climax
決意	Ketsui — resolve	passionate	Strength of will
喜び	Yorokobi — joy	positive	Elation / finale
華やかさ	Hanayakasa — splendor	positive	Glittering elation
軽快	Keikai — lightness	positive	Good tempo, brisk
緊張	Kinchō — tension	tense	Dramatic tautness

The same family (e.g. 哀愁⇔郷愁⇔諦念) connects naturally; a sudden jump across families needs a bridge track, which the engine accounts for. When unsure, choose by "what the dancer feels in this track". Enter the Japanese term in the cell.

4. flow_type (the track's shape)

Value	Meaning	Example
`Descent`	Descending. From peak to calm	Tail-fade type (Bahía Blanca-like)
`Ascent`	Ascending. From calm to heat	Drama type (Adiós Nonino-like)
`Sustain`	Sustained, steady energy	Continuous tension (La Yumba-like)
`Bounce`	Bouncing, undulating	Catchy type (El Choclo-like)

5. intensity (1–5)

Value	Level	Feel
1	Ambient	Atmospheric, light intro
2	Low	Calm, moderate
3	Medium	Standard tango
4	High	Builds up
5	Max Peak	Climax, finale

6. Example (one track)

Column	Value
title	Bahía Blanca
orchestra	Carlos Di Sarli
genre	tango
year	1957
singer	(blank = instrumental)
primary_mood	哀愁
secondary_moods	郷愁, 切望
flow_type	Descent
intensity	3
notes	Downer masterpiece / for opening or closing

7. When a track won't match (matching tips)

On import the app searches in this order:

track_id matches
Title matches + orchestra loosely matches (Biagi and Rodolfo Biagi both OK)
If the title is unique in the library, match by that

Case, full/half-width and bracket notation like [Tango] are absorbed automatically. If it still says "No matching track":

Check that title/orchestra spelling isn't very different from the library (especially feat. notation)
Most reliable: base your file on the CSV from ⬆ Export tags (it has track_id, so it always matches)
If multiple same-titled tracks exist and orchestra can't disambiguate, it is intentionally skipped to avoid wrong matches

8. Don'ts & cautions

Don't change the header spelling (row 1) or delete columns
Don't leave primary_mood blank and assume it's "tagged" (blank rows aren't synced)
Don't break formatting by "saving over" the CSV → for importing, save as .xlsx
Don't keep the Excel in only one place → back it up often (it is the source of truth)

Import = "Excel wins / blanks are left alone". Importing with blank columns won't erase existing tags, but a cell where you did put a value will overwrite. Review once before importing.

9. Recommended approach

Start with the orchestras you use most (your standard tandas)
Copy-paste primary_mood across same orchestra/era at once
Once the 3 required are done, flesh out secondary_moods and year/singer
While building tandas live, note tracks where a suggestion "felt wrong" → reflect in Excel
Use the report's "still-untagged tracks" as your to-do list for next time