Skip to content
RU
All projects

music-analyze

Merges Yandex Music and SoundCloud likes into one list, cleans up re-uploads and ranks artists, albums and genres.

Discuss a similar project

Search “tramvai” or “lagoon”, roll a random track, open a row

everything you liked on Yandex Music and SoundCloud, piled up and counted

Tracks
4,255
Likes
4,684 before merge
Duplicates merged
429
Artists
1,585
Genres
246
Hours of music
226
a random track from your likesБетонный рассветПолярная Ночь · on both platforms · liked in 2023
/

Who’s on top

By number of liked tracks. Click a row to spill all its tracks.

PlatformYear

Which albums

By number of liked tracks from the album. The album is known for 81% of tracks: Yandex almost always has it, SoundCloud only when the uploader kept the tags. 502 albums have more than one liked track.

PlatformYear

What you live on

Tags from MusicBrainz and Last.fm, not whatever the uploader typed.

CountYear

How the taste changed

Each track counts once, by the year you first liked it. Click a column to filter everything above.

2023202420252026

2023639 tr.

  1. 1Tin Lagoon18
  2. 2Moth Lantern15
  3. 3Velvet Static12
  4. 4Полярная Ночь10
  5. 5Hollow Saints9
  6. 6Wraith Garden8

2024812 tr.

  1. 1Полярная Ночь14
  2. 2Moth Lantern14
  3. 3Tin Lagoon12
  4. 4Velvet Static10
  5. 5Северный Ветер9
  6. 6Kestrel Unit9

2025851 tr.

  1. 1Полярная Ночь16
  2. 2Tidepool Radio14
  3. 3Северный Ветер12
  4. 4Kestrel Unit11
  5. 5Tin Lagoon10
  6. 6Лунный Трамвай9

20261,953 tr.

  1. 1Северный Ветер26
  2. 2Полярная Ночь24
  3. 3Лунный Трамвай21
  4. 4Tidepool Radio20
  5. 5Kestrel Unit18
  6. 6Tin Lagoon18

What lives where

How many tracks live on one platform only, and how many on both.

SoundCloud only983
On both349
Yandex only2,923

How it’s counted

Where the data comes from

Likes are taken straight from a logged-in session: SoundCloud via api-v2, Yandex via api.music.yandex.ru. No passwords or tokens are stored.

Cleaning re-uploads

[FREE DL], (prod. by …), @mentions, #tags and track numbers are stripped from titles. The artist comes from ID3, then from the “Artist — Track” pattern, and only last from the account name.

Restoring authorship

If a track was still credited to a blog, its title was checked against a dictionary of artists from trusted metadata, by whole words, or “ROT” would grab “Rotation”.

Merging platforms

One track if the title is close, an artist matches and the length differs by 7 seconds at most. “Полярная Ночь” and “POLYARNAYA NOCH” are one artist. A remix stays a separate track.

Genres

Tags from MusicBrainz and Last.fm by canonical name; Deezer and iTunes as backup. The platform’s own “genre” comes last.

Colours

Bar colours are just stickers. Length and the number beside it carry the value, not colour.

A mini copy of the report. Totals are real, the lists are a demo
My role
Data collection, cleaning, report
Year
2026
Likes
4,684
Duplicates merged
429

What it does

  • Likes from two platforms in one list
  • Duplicate tracks are merged
  • Genres from MusicBrainz and Last.fm
  • A report with search and filters

How it’s built

  • Python, no third-party libraries
  • Junk in titles is removed by rules, not by hand
  • External service responses are cached

Worth knowing

  • A personal tool, not a service
  • Yandex is read from an open browser session
  • No repository yet
Next projectQuotrum

Discuss a similar project

Need something like this? Write to me and I’ll tell you how long it will take.

Message on Telegrams001ar.dev@gmail.com