Listening Notes - Suno vs ACE 1.5 (2026-08-17)

Track: "The Hand That Stays Out"

Objective Measurements

|--------|------|---------| | Duration | 3:05 (184.6s) | 3:22 (202.0s) | | Integrated loudness | -13.5 LUFS | -14.5 LUFS | | True peak | -0.7 dBTP | -1.2 dBTP | | Loudness range (LRA) | 6.1 LU | 5.3 LU | | Overall RMS | -15.2 dB | -16.1 dB | | Peak | -0.7 dB | -1.3 dB |

Section-by-Section RMS (10s windows)

|------|----------|------------|-------| | 0s | -19.5 | -19.8 | Both start quiet - intro/verse 1 | | 10s | -18.0 | -17.7 | Both rising - verse 1 continues | | 40s | -16.8 | -17.2 | Verse 1 → chorus 1 transition | | 60s | -15.6 | -16.1 | | | 80s | -16.3 | -16.1 | | | 90s | -15.9 | -19.5 | ACE drops - possible verse 2 / quieter section | | 150s | -14.3 | -14.1 | Both in outro range | | 160s | -15.8 | -14.7 | ACE slightly louder (outro continuation) | | 180s | - | -14.8 | ACE only (longer track) | | 190s | - | -18.5 | ACE final fade |

Key Observations

The Turn at ~110s is clearly visible - RMS drops to -23.1 dB, the lowest point in the track. The bed genuinely drops out.
Chorus 2 at 130-140s is the loudest section (-14.0 to -13.7 dB) - the B4 climax on "stays out" is real.
The dynamic architecture matches the plan: flat verses (-15 to -17), Turn drop (-23), C2 peak (-13.7), outro (-12.5 at the very end).
The -23.1 dB drop at 110s is a 3x RMS cliff from the verse level (~-15.6) - exactly the breathturn the composition plan specified.

ACE 1.5 track:

More consistent level overall - less dramatic dynamic shifts.
The track is 17 seconds longer (202 vs 184.6s) - ACE may be rendering at a slightly different tempo or extending sections.
The outro fades more gradually (170s -15.6 → 190s -18.5).

Gemma4:e2b Listening (verse/clip section)

Detected: acoustic guitar, clear intimate vocal, narrative/conversational style
Vocal: front and center, gentle and reflective
Tempo: moderate to slow
Mood: warm, nostalgic, reflective
Arrangement: sparse, focused on guitar + vocal
Production: warm, natural, clean, intimate
Genre: Folk / Acoustic Singer-Songwriter

ACE 1.5 (55-85s clip):

Detected: acoustic guitar, warm conversational vocal
Similar folk/singer-songwriter characterisation
Notable: gemma flagged "organic imperfections" and "human warmth" as differentiators from typical AI music
Described as having "genuine human warmth" that AI models struggle to replicate

Comparison Summary

1. Both tracks are in the same genre - folk/singer-songwriter with acoustic guitar and intimate vocal. The composition translated well to both renderers.

3. ACE is more consistent - less variation between sections, gentler transitions. This could be a strength (consistency) or a weakness (less emotional impact from the breathturn).

4. ACE is longer - 17 seconds longer. Either the tempo is slightly slower or sections are extended. This needs investigation.

5. ACE has no Dakota voice - the vocal is present but generic. Once we train the voice model, this becomes the differentiator.

Next Steps

Listen to both full tracks to validate the RMS section analysis
Compare ACE tempo vs the 118 BPM plan (is it rendering at a different speed?)
When voice training is ready, re-render with Dakota's voice
Consider ACE as the primary production path for songs where arrangement control matters