Music v2 Benchmark Part 2: ElevenLabs vs. Suno v5.5 Genre Shifts

We put the genre-shifting capabilities of ElevenLabs Music v2 to the test against Suno v5.5, experimenting with both vocal tracks (Opera to Heavy Metal) and instrumentals (Orchestra to Metal). While ElevenLabs v2 provides clear genre separation, it struggles with structural stability in vocal transitions. Conversely, Suno produces high-quality results but tends to blend the two styles into a hybrid. In the instrumental category, ElevenLabs emerges as the clear winner with superior stylistic precision.

"From opera to heavy metal, seamlessly within a single track."
We put ElevenLabs Music v2’s genre-shifting capabilities to the ultimate test, going head-to-head against Suno v5.5 across both vocal and instrumental tracks!

 

Welcome to the Sonetho. ⚡

In our previous post, we compared v1, v2, and Suno using identical lyrics and genres, concluding that while "v1 to v2 is a clear step up, there remains a gap compared to Suno."
This time, we’re tackling a much tougher challenge — the ability to switch genres within a single song.

One of the highlights of the ElevenLabs Music v2 launch was the promise of a "seamless transition from opera to heavy metal."
We wanted to see if that’s truly possible and how Suno v5.5 handles the same request — so we ran two experiments to find out.

The bottom line: Both models exhibited distinct strengths and limitations when faced with this complex task.
Eleven v2 clearly distinguishes between genres, but stability often wavered at the point of transition. Suno delivers high production value but tends to blend the two styles together rather than switching between them.

 

🧪 Experimental Setup

  • Experiment 1: Vocal Track — Opera (Bel Canto) → Heavy Metal.
    We evaluated whether the vocals successfully shift from an operatic tone to distorted screaming or powerful rock vocals, and if the intensity of the composition syncs with the vocal shift.

  • Experiment 2: Instrumental Track — Orchestra → Metal band (no vocals). We focused strictly on stem separation and the precision of the transition.

  • Models used: Eleven Music v2, Suno v5.5

  • Mode: Custom (Manual lyrics and prompt input, identical conditions)

 

🎤 Experiment 1: Vocal Track — Can Opera become Heavy Metal?

 

Style Prompt

Cinematic grand opera shifting into heavy metal. Dramatic classical orchestration, symphonic opera soprano vocals, sudden intense mid-track transition, heavy distorted electric guitar riffs, thunderous double-bass drums, aggressive vocal tone change, high-fidelity studio mix.

 

Lyrics (using structure tags to mark the transition point)

[Intro - Opera]
(Grand orchestral strings fade in)
(Soprano warming up: Ah-ah-ah)

[Verse - Opera Style]
The curtain rises, the stage is set
An ancient sorrow, a grand regret
O, fate divine, hear my lonely cry!
(Symphonic choir rising in background)

[Pre-Chorus - Rising Opera Climax]
The lights are fading, the shadows grow long
This is the end of the beautiful song...
(High opera belt, holding the final note)

[Sudden Metal Transition]
[Heavy Metal Drop]
(Boom! Heavy guitar distortion crash, aggressive blast beats)

[Chorus - Heavy Metal Style]
Now smash the walls! Break the chains!
Feel the fire runnin' through your veins!
No more silence, scream it out loud!
We are the chaos, we rule the crowd!

[Outro]
(Screaming: Yeah! Burn it down!)
(Heavy guitar solo riff chugging)
(Sudden hard stop)

 

① Eleven Music v2 (1st generation)

 

② Eleven Music v2 (2nd generation)

 

③ Eleven Music v2 (3rd generation)

 

👍 What went well

  • Genre distinction is clear. The opera section features authentic operatic instrumentation (strings, choir), and the vocals hit those soprano notes effectively.

  • Vocal quality: While the reverb varies with each output, the opera segments are quite listenable.

👎 Areas for improvement

  • Melodic development is somewhat flat, with similar melodies repeating.

  • The key observation — stability dropped noticeably during the heavy metal transition.
    The tempo, instrumentation, and vocals lost cohesion, making it hard to hold the song's shape. This happened across all three attempts.

(My ElevenLabs credits, ouch! Unlike Suno, which we only ran once, v2 failed to produce a stable result even after three tries.)

 

④ Suno v5.5

(We generated this once, got two tracks, and uploaded one without any re-rolling.)

 

👍 What went well

  • The overall polish is high. The track sounds fully produced in terms of tempo, mixing, and vocal performance.

  • Reliability: One generation provided a stable, coherent result without the model breaking down.

👎 Areas for improvement

  • Genre distinction is blurred.
    Instead of a distinct shift, it creates a "Symphonic Metal" hybrid where the two genres are mashed together from the start.

  • Even though we specified a "Sudden Metal Transition," the model interprets it as a "steady fusion" rather than a sharp change.

 

🎼 Experiment 2: Instrumental Track — Genre shifting via instrumentation only

This time, we removed the vocal variable to compare how the models handle build-ups, drops, and instrument swaps. This is a true test of v2’s "stem separation" and transitional prowess.

 

Style Prompt

Instrumental orchestral opera transitioning into aggressive progressive metal, No vocals, classical string ensemble, solo dramatic cello, sudden explosive dynamic shift, down-tuned 8-string electric guitars, complex heavy drum polyrhythms, orchestral elements merging with metal riffs, crystal clear instrument separation, pristine master.

(No lyrics input)

 

⑤ Eleven Music v2

Wow... there is a distinct transition where the genre shifts mid-track.
The orchestra builds up, and then there is a clear, definitive cut to the metal session. I was genuinely impressed; Eleven Music definitely holds its own in the BGM space.

 

⑥ Suno v5.5

Again, it’s a hybrid. It turns into a "Symphony Rock" track where the orchestra and rock elements are permanently fused...!
The quality is great, but there is no actual "shift."

 

📊 Final Verdict

🎯 Distinct Characteristics

Eleven Music v2: "The ambitious model that tries to cleanly separate genres." Its ability to delineate genres is clearly superior to the competition. However, it struggles to maintain model stability during vocal-heavy transitions. It’s still a work in progress—it needs more tuning to reach Suno v5.5’s level of production consistency.

Suno v5.5: "The highly polished, safe hybrid model." The output is always refined and radio-ready. However, even when asked for a "sudden transition," it defaults to a smooth, safe blend. It’s sonically secure, but lacks the experimental edge for radical genre-switching.

🎼 Potential in the BGM/Instrumental Market

Eleven Music v2 truly shines in instrumental transitions. Without the complexity of vocals, the model stays stable and executes the genre swap exactly as requested. This is a strong signal that Eleven Music is highly competitive in the BGM, gaming, and film soundtrack spaces.

🔮 What’s next?

v2’s focus on "genre separation" is a meaningful, necessary direction. Moving away from the "mash-up" approach toward "distinct, sequential switching" is harder, but it opens the door to more musically interesting results. Once they solve the stability drop during vocal transitions—whether in v3 or a 2.5 update—this is going to be a powerhouse model.

 

🎵 Want to try switching genres yourself?

If this comparison piqued your curiosity, try switching genres yourself with ElevenLabs Music v2. New sign-ups get 50% off their first month, so you can dive right in.

Create with ElevenLabs Music v2 →

🔮 Upcoming Experiments

Next time, we’ll dive into another feature ElevenLabs has been highlighting — Stem Editing. We’ll see how cleanly we can isolate and remix vocals, drums, guitars, and synths, and compare the result against Suno’s own stem capabilities.

Stay tuned! 🎬

 

📚 Recommended Reading

 

Give these tracks a listen and let us know in the comments: "Which model made a more meaningful attempt at the genre-shift challenge?" 🎵
Sonetho ⚡