Try AnyToSpeech
Benchmark › Music models › Stable Audio 2.5
Stability AI · tested 2026-11-22

Stable Audio 2.5

93.5overall · rank 1 of 8 music models
API model: stable-audio-2.5 Reached through Replicate 44.1 kHz stereoInstrumental onlyControls: Length field

Profile

Each spoke is one test, 0 at the centre and 100 at the edge.

Stable Audio 2.5Median of all models (dashed)

Test by test

The dot shows where this model sits among all 8: left is last, right is first.

Instrumental briefs

Each brief asked for a genre, an exact tempo, no vocals, and a 30 second track. One square per brief for each test; hover for details.

Genre
21/24
Tempo
22/24
Length
24/24
No vocals
24/24
Clean ending
23/24

Listen

Previews are the first 20 seconds.

lo-fi hip hop, 80 BPMmeasured 80 BPM · 30 s long
genre ✓tempo ✓length ✓
country, 95 BPMmeasured 191 BPM · 30 s long
genre ✓tempo ✓length ✓
funk, 110 BPMmeasured 110 BPM · 30 s long
genre ✓tempo ✓length ✓
techno, 130 BPMmeasured 130 BPM · 30 s long
genre ✓tempo ✓length ✓
punk rock, 160 BPMmeasured 80 BPM · 30 s long
genre ✓tempo ✓length ✓
drum and bass, 174 BPMmeasured 116 BPM · 30 s long
genre ✓tempo ✓length ✓
reggae, 75 BPMmeasured 75 BPM · 30 s long
genre ✗tempo ✓length ✓
disco, 120 BPMmeasured 120 BPM · 30 s long
genre ✓tempo ✓length ✓
heavy metal, 150 BPMmeasured 75 BPM · 30 s long
genre ✓tempo ✓length ✓
jazz, 140 BPMmeasured 141 BPM · 30 s long
genre ✓tempo ✓length ✓
bossa nova, 100 BPMmeasured 67.5 BPM · 30 s long
genre ✓tempo ✗length ✓
trap, 150 BPMmeasured 75 BPM · 30 s long
genre ✓tempo ✓length ✓

Songs with supplied lyrics

This model is instrumental only, so the lyric test does not apply and its overall score is built from the tests it supports.

Compare with

Think a score is wrong? Check it.

Every test item, scoring rule and known limit is published, and the audio behind each number is on the page.

Read the methodology