Music Descriptor Module 3
Input: 313-d = 41-d M1 GT labels + 16-d M2 GT labels + 256-d M2 context_vector. Final module in the Scene → Narrative → Music pipeline.
8 Music Descriptor Heads
| # | Head | Type | Output |
|---|---|---|---|
| 1 | tempo_bpm | regression | 45–170 BPM |
| 2 | musical_valence | regression | -1.0 to 1.0 |
| 3 | tonality | 3-class | atonal, major, minor |
| 4 | harmonic_style | 7-class | atonal, chromatic, cluster, diatonic, modal, pentatonic, whole_tone |
| 5 | dynamic_shape_m4 | 8-class | crescendo, diminuendo, flat, subito_forte, subito_piano, sustained, swell, terraced |
| 6 | rhythm_style | 6-class | drive, off, ostinato, pulse, rubato, sparse |
| 7 | texture | 5-class | ambient, chamber, full, hybrid, solo |
| 8 | orchestration | 14-label | ambient_pad, brass, choir, electronic, ethnic, guitar, harp, organ, percussion, piano, solo_voice, strings, synth, woodwinds |
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support