AI Music Candidate Comparison Worksheet: Choose the Best Generation
Selection worksheet
Choose the strongest foundation—not merely the cleanest generation.
Score each candidate against the same project brief so novelty, preference and credit fatigue do not become the only criteria.
1. Label the candidates
Record the output link, file name, tool, model, date, prompt version and uploaded sources. Use no more than five finalists in one round.
2. Score five criteria from 1–5
| Criterion | 1 | 3 | 5 |
|---|---|---|---|
| Emotional impact | Little connection | Some strong moments | Immediate and sustained effect |
| Song identity | Generic or confused | Recognizable direction | Distinctive and coherent |
| Vocal credibility | Distracting or wrong | Usable with repairs | Believable and aligned |
| Structural strength | Sections do not build | Basic structure works | Strong arrival and resolution |
| Edit potential | Near-total rebuild | Several fixable weaknesses | Strong base with targeted changes |
3. Complete the score sheet
| Candidate | Emotion | Identity | Vocal | Structure | Edit | Total /25 |
|---|---|---|---|---|---|---|
| A | __ | __ | __ | __ | __ | __ |
| B | __ | __ | __ | __ | __ | __ |
| C | __ | __ | __ | __ | __ | __ |
| D | __ | __ | __ | __ | __ | __ |
21–25: strong candidate. 16–20: promising but requires work. 11–15: useful reference or partial salvage. 10 or lower: archive and move on. This is a decision aid, not an objective quality score.
4. Choose the next operation
Write the decision
Candidate ___ is selected because it best delivers ___, while the first production priority is to fix ___.
Save rejected candidates and rejection reasons to prevent circular decisions and support the contribution file.