Symbolic, not sonic
Only textual pitch and duration sequences reach the model.
MeloText-ID · Cross-representational AI benchmark
Can an AI hear music in plain text?
A melody enters as symbols. No audio. No tools. Twenty-five possible songs. MeloText-ID measures whether a language model can recover the tune—even when rhythm disappears or every pitch moves to a new key.
“Hear” is shorthand for successful cross-representational inference, not a claim about subjective experience.
Demonstration only · not scored
What do you hear?
C C G G A A G | F F E E D D C
Verified, not judged
Exact choice IDs are graded automatically. No LLM evaluator.
Transposed, not copied
Key changes test identity beyond absolute-pitch matching.
Highest official score41.36%Gemini 3.6 Flash
Closed-set chance4%1 of 25 candidates
Best / chance10.3×relative lift
10,000 scored decisions10 complete models25 melodies × 4 conditions✓ Audit passed
Interactive results
Choose what the ranking means.
The official score includes all four conditions. Select any non-empty combination to inspect a different capability slice.
| Rank | Model | Score | vs chance | 10-run range | Cost | Details |
|---|---|---|---|---|---|---|
| #1 | Gemini 3.6 Flashgoogle/gemini-3.6-flash | 41.36% | 10.3× | 37–43% | $2.75 | 10 runsProvider pin: google-ai-studio Arithmetic mean: 41.40%
0 parse failures · 0 infrastructure failures |
| #2 | GPT-5.6 Solopenai/gpt-5.6-sol | 40.50% | 10.1× | 36–45% | $13.15 | 10 runsProvider pin: openai Arithmetic mean: 40.60%
0 parse failures · 0 infrastructure failures |
| #3 | Claude Opus 4.8anthropic/claude-opus-4.8 | 35.36% | 8.8× | 34–40% | $5.42 | 10 runsProvider pin: anthropic Arithmetic mean: 35.40%
0 parse failures · 0 infrastructure failures |
| #4 | Kimi K3moonshotai/kimi-k3 | 31.59% | 7.9× | 29–37% | $9.33 | 10 runsProvider pin: moonshotai/int4 Arithmetic mean: 31.70%
34 parse failures · 0 infrastructure failures |
| #5 | Claude Sonnet 5anthropic/claude-sonnet-5 | 22.51% | 5.6× | 19–26% | $1.90 | 10 runsProvider pin: anthropic Arithmetic mean: 22.60%
0 parse failures · 0 infrastructure failures |
| #6 | GPT-5.6 Lunaopenai/gpt-5.6-luna | 22.34% | 5.6× | 20–26% | $3.85 | 10 runsProvider pin: openai Arithmetic mean: 22.40%
16 parse failures · 0 infrastructure failures |
| #7 | Qwen 3.7 Maxqwen/qwen3.7-max | 20.12% | 5.0× | 18–24% | $1.02 | 10 runsProvider pin: alibaba Arithmetic mean: 20.20%
2 parse failures · 0 infrastructure failures |
| #8 | GLM 5.2z-ai/glm-5.2 | 14.98% | 3.7× | 12–18% | $0.83 | 10 runsProvider pin: z-ai/fp8 Arithmetic mean: 15.10%
0 parse failures · 0 infrastructure failures |
| #9 | DeepSeek V4 Prodeepseek/deepseek-v4-pro | 10.06% | 2.5× | 6–14% | $0.68 | 10 runsProvider pin: ionstream/fp4 Arithmetic mean: 10.40%
0 parse failures · 0 infrastructure failures |
| #10 | MiniMax M3minimax/minimax-m3 | 4.99% | 1.2× | 2–11% | $0.19 | 10 runsProvider pin: minimax/fp8 Arithmetic mean: 5.50%
10 parse failures · 0 infrastructure failures |
Bars use a fixed 0–100% scale. The vertical mark is 4% chance. A zero geometric score can mean one repetition scored zero; no epsilon is added.
Four ways to recognize one tune
Condition matrix
Absolute geometric scores. Select a condition cell to move the leaderboard to that slice.
| Model | Overall | Original · Notes only | Original · Pitch + duration | Transposed · Notes only | Transposed · Pitch + duration |
|---|---|---|---|---|---|
| Gemini 3.6 Flash | 41.36% | ||||
| GPT-5.6 Sol | 40.50% | ||||
| Claude Opus 4.8 | 35.36% | ||||
| Kimi K3 | 31.59% | ||||
| Claude Sonnet 5 | 22.51% | ||||
| GPT-5.6 Luna | 22.34% | ||||
| Qwen 3.7 Max | 20.12% | ||||
| GLM 5.2 | 14.98% | ||||
| DeepSeek V4 Pro | 10.06% | ||||
| MiniMax M3 | 4.99% |
Darker cells indicate higher scores · every cell is labeled · chance is 4%
Why it matters
Intelligence should travel between senses.
Humans can look at a silent line of notation and experience the shape of a melody. MeloText-ID asks whether language models can perform an analogous representational crossing.
A note string and a melody concept begin in different regions of representational space. Recognition requires a mapping that preserves interval relationships, rhythmic contour, and identity as the key changes.
Structural reconstruction
Tokens become an ordered melodic contour.
Temporal binding
Duration codes add rhythmic structure.
Invariant identity
Transposition changes pitch, not the tune.
Research context
These papers motivate the behavioral question; they are not evidence about any model's internal mechanism.
From score to controlled test
How the benchmark was built
- 01
Choose
25 highly familiar, nameable melodies with clear monophonic motifs.
- 02
Extract
The most recognizable passage, checked against score sources.
- 03
Encode
Notes-only and duration-prefixed strings in 32nd-note units.
- 04
Transpose
Move the key while preserving intervals, attacks and measures.
- 05
Randomize
25 titles per item, with every correct answer position balanced.
- 06
Repeat
Ten independently randomized 100-item runs, graded exactly.
One melody, two textual surfaces
Notes onlyC C G G A A G
Pitch + duration8C 8C 8G 8G 8A 8A 16G
Task, prompt and stimulus construction
Every stateless request contains one symbolic melody and all 25 candidate titles. Models receive no audio, browsing, tools, metadata, or cross-request context. `R` means rest; accidentals and octave markers are encoded explicitly.
Download the frozen prompt ↓Randomization, balance and grading
Candidate permutations are deterministic. Within each condition and repetition, every answer position is correct exactly once. Exact JSON choice IDs are graded automatically; malformed fully received answers remain incorrect.
Scoring and aggregation
Within each repetition, accuracy is arithmetic. Across ten repetitions, the score is the exact geometric mean. For a custom slice, selected cells are pooled inside each repetition before taking the geometric mean. No epsilon is added.
score = (A₁ × A₂ × … × A₁₀)^(1/10)
Model routing, costs and audit
Exact model/provider routes were pinned and fallback was disabled. The complete scored panel cost $39.111479. Hash, route, balance, scoring, export and cost checks passed.
Provenance and reproducibility
Version 0.1.3-pilot · finalized July 23, 2026 · source run live-20260722T122046Z-a0d2fe0276.
Experiment fingerprint: a0d2fe02767f66b48a6cce5a993698b44786efb78aab917b27f73225a2ef385c
The recognition set
25 melodies people remember.
The pilot favors recognizable motifs and reproducible monophonic encodings. It is not a balanced survey of global musical culture.
- 011812 Overture
- 02Ave Maria (Schubert)
- 03Beethoven's Fifth Symphony
- 04Ode to Joy (Beethoven's Ninth Symphony)
- 05Bella Ciao
- 06The Blue Danube
- 07Canon in D (Pachelbel)
- 08Clair de Lune
- 09Eine kleine Nachtmusik
- 10Fur Elise
- 11Hallelujah Chorus
- 12Happy Birthday to You
- 13In the Hall of the Mountain King
- 14The Internationale
- 15La donna e mobile
- 16March of the Volunteers (Chinese National Anthem)
- 17Mo Li Hua (Jasmine Flower)
- 18O Sole Mio
- 19Pomp and Circumstance March No. 1
- 20Queen of the Night Aria
- 21Rondo Alla Turca (Turkish March)
- 22The Star-Spangled Banner
- 23Toreador Song
- 24Two Tigers (Frere Jacques melody)
- 25Spring (Vivaldi, The Four Seasons)
Read the result precisely
Provenance & limitations
What this benchmark shows
Behavior consistent with cross-format musical reconstruction: parsing symbolic notation, preserving relational structure, and mapping it to a familiar semantic identity.
Cultural and training-data confounds
The small set is weighted toward Western classical music. Cultural familiarity and musical training matter, and familiar note sequences may have appeared in pretraining data.
What the representation omits
Monophonic text removes harmony, timbre, dynamics, articulation and performance. Closed-set recognition is not melody generation or general musical ability.
Why is Grok absent?
Grok 4.5 was stopped before completion after network instability and excluded entirely. No partial Grok result contributes to any rank or aggregate.
Official versus exploratory
The all-condition leaderboard is the official result. Every subset ranking is a transparent diagnostic view and should not replace the preregistered score.
Human comparison
This pilot contains no human baseline and cannot establish subjective auditory imagery, literal synesthesia, consciousness, or the internal method used by a model.
Open aggregate evidence
Inspect the benchmark.
Public files contain aggregate scores, protocol and integrity commitments—never raw answers or an answer key.
Independent research
Created by
Alejandro Zarzuelo.
MeloText-ID was conceived and created to investigate how AI systems connect symbolic structure, imagined sensory form, and semantic identity.
Visit alejandrozarzuelo.com ↗