Streaming & Charts

Do Club Remixes Actually Hit Harder Than the Original?

Post 01 · Track Record · Dataset: Spotify audio features (TidyTuesday, 2020)

You know the moment: a song you already like comes back six months later as a "Remix," suddenly louder, faster, more built for a dance floor. It feels more intense. But is that actually true, according to Spotify's own numbers — or is it just louder mixing and a new drop?

Slope chart comparing energy of original vs remix tracks, and a bar chart comparing averages

The dataset

This uses Spotify's audio-feature data (danceability, energy, tempo, valence, etc.) as published in the TidyTuesday "Spotify Songs" dataset (2020-01-21), originally pulled via the Spotify API and released under a public/CC0 license. The full dataset has 32,833 tracks.

For this post I worked from a 40-track sample pulled from three Spotify "Pop Remix," "Dance Pop," and "Dance Room" playlists — chosen specifically because several songs show up in both their original and remixed form, which is what makes a real comparison possible. You can download exactly what I used below, or grab the full dataset from the link above to run your own version.

Downloads

spotify_songs_sample.csv — the 40-track working sample
post01_matched_pairs.csv — the 7 matched original/remix pairs

The method

Using Python (pandas + matplotlib):

Flagged any track with "Remix" in its title.
Stripped remix suffixes to recover a "base title," then grouped by artist + base title to find songs with both an original and at least one remix in the sample.
Compared Spotify's energy score (0–1, how intense/fast a track feels) between each matched pair.
Charted the shift per song (slope chart) and the overall group averages across all 40 tracks.

df["is_remix"] = df["track_name"].str.contains("remix", case=False)

def base_title(name):
    name = re.sub(r"\s*-\s*.*remix.*", "", name, flags=re.IGNORECASE)
    name = re.sub(r"\s*\[.*remix.*\]", "", name, flags=re.IGNORECASE)
    return name.strip()

df["base_title"] = df["track_name"].apply(base_title)
# group by (artist, base_title) to find original/remix matches
# then compare energy: remix_energy - orig_energy

Full script: scripts/post01_remix_energy.py (included in the project files).

What the matched pairs showed

Song	Remix	Original energy	Remix energy	Change
Post Malone (feat. RANI) — Sam Feldt	GATTÜSO Remix	0.642	0.903	+0.261
Call You Mine — The Chainsmokers	Keanu Silva Remix	0.702	0.930	+0.228
Stay (Don't Go Away) — David Guetta	Mark Knight Remix	0.797	0.995	+0.198
Call You Mine — The Chainsmokers	Asketa & Natan Chaim Remix	0.702	0.844	+0.142
Crash Into Me — Steve Aoki	Settle Down Steavis Aoki Remix	0.815	0.956	+0.141
Never Really Over — Katy Perry	R3HAB Remix	0.882	0.856	−0.026
With You — Kaskade	LöKii Remix	0.921	0.727	−0.194

5 of the 7 matched pairs got a higher energy score in remix form — sometimes by a lot (Post Malone/Sam Feldt jumped +0.26). But two went the other way: Kaskade's "With You" is actually calmer as the LöKii remix than the original. Across the full 40-track sample, remixes averaged 0.840 energy vs. 0.763 for originals — a real but not universal bump.

Takeaway: "Remix" leans toward more energy, but it's a trend, not a rule — about 1 in 4 matched remixes in this sample actually dialed things down. And with only 7 matched pairs, this is a fun signal, not a verdict — the full 32,833-track dataset would be the place to check if this holds at scale.

← Back to all posts