Sound Quality
25%Audio quality, musicality, vocal and arrangement coherence, artifacts, and consistency.
Methodology · V1.1
How we score AI music tools.
Product quality and evidence confidence answer different questions. We publish both, keep popularity separate, and never let commercial relationships change rank.
01 · Product score
Six fixed dimensions, weighted to 100%
Reviewers score each dimension from 0 to 100 against written anchors. The weighted result is retained to at least four decimal places for ordering and shown to one decimal place.
Creative Control
20%Prompt, structure, style, iteration, stems, and local editing control.
Workflow & Usability
15%Learning curve, generation speed, reliability, export, and project workflow.
Value
15%Free access, effective cost, credit clarity, and capability at the same price.
Rights & Transparency
15%Commercial use, ownership, restrictions, terms clarity, and disclosure quality.
Reliability & Maturity
10%Availability, update history, support, platform stability, and operating continuity.
Overall score
Sound × .25 + Control × .20 + Workflow × .15 + Value × .15 + Rights × .15 + Reliability × .1002 · Evidence confidence
Confidence never boosts the product score
Confidence describes how complete, independent, current, and internally consistent the evidence is. A promising tool with weak evidence remains unranked instead of receiving a precise-looking estimate.
Coverage
All six dimensions meet their minimum evidence gate.
Independence
Quality claims are not supported only by the vendor.
Freshness
High-risk facts such as pricing and rights are current.
Source quality
Primary and reputable sources carry more evidentiary weight.
Consistency
Material conflicts are resolved or disclosed, never silently averaged.
03 · Ranking eligibility
A score is withheld until the publication gates pass
- All six dimension scores are valid.
- Sound, control, and workflow each trace to at least two current, independent A/B evidence records.
- The quality evidence spans at least two source families and includes disclosed product use.
- Evidence confidence is at least medium.
- Material pricing, rights, and availability conflicts are resolved or disclosed.
- An editor approves the score version before activation.
Tied four-decimal scores share a rank. The next position uses competition ranking, and a tool slug is used only for stable display order—not as a quality tie-breaker.
04 · Evidence layers
Facts, reviews, and context play different roles
- Official factsFeatures, pricing, limits, availability, commercial terms, and release history.
- Independent use evidenceCurrent professional tests and actual-use reviews, graded for method, recency, independence, and commercial disclosure.
- Corroborating contextReview aggregates and recurring community patterns can reveal themes, but cannot replace the independent evidence minimum.
- Optional AIMUSIC.GUIDE testingFirst-party tests may investigate a conflict or calibrate the rubric, but are not a universal requirement and cannot bypass missing public evidence.
05 · Regional context
One ranking, with regional limits kept beside the facts
Sound, creative control, and workflow evidence describe the product itself. Availability, price, free access, taxes, commercial terms, and country restrictions retain their evidence scope. We show those limits directly without manufacturing a second score or leaderboard.
06 · Limitations and changes
A ranking is a versioned editorial judgment, not permanent truth
AI music products change quickly. Every published ranking identifies its score version and publication date. Material corrections create a new reviewed revision; old snapshots remain auditable and can be restored if a release is wrong.