File size: 5,052 Bytes
d69b6f3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eebdf45
d69b6f3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cd22718
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
55dc181
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
---
tags:
- model_hub_mixin
- pytorch_model_hub_mixin
- audio
- rhythm-game
- music
---

# GameChartEvaluator (GCE4)

A neural network model for evaluating the quality of rhythm game charts relative to their corresponding music. The model predicts a quality score (0-1) indicating how well a chart synchronizes with the music.

## Model Architecture

The model uses an early fusion approach with dilated convolutions for temporal analysis:

1. **Early Fusion**: Concatenates music and chart mel spectrograms along the channel dimension (80 + 80 = 160 channels)
2. **Dilated Residual Encoder**: 4 residual blocks with increasing dilation rates (1, 2, 4, 8) to capture multi-scale temporal context while preserving 11ms frame resolution. This gives the model a **receptive field of ~0.73s** (63 frames), meaning each time-step's score depends on the local ~0.36s context before and after.
3. **Error-Sensitive Scoring Head**: Combines average local scores with the worst 10% of scores using a learnable mixing parameter

```
Input: (B, 80, T) music_mels + (B, 80, T) chart_mels
  ↓ Concatenate
(B, 160, T)
  ↓ Conv1D Projection
(B, 128, T)
  ↓ Dilated ResBlocks × 4
(B, 128, T)
  ↓ Linear → Sigmoid (per-frame scores)
(B, T, 1)
  ↓ Error-Sensitive Pooling
(B,) final score
```

## Usage

```python
import torch
from gce4 import GameChartEvaluator

model = GameChartEvaluator.from_pretrained("JacobLinCool/gce4")
model.eval()

# Input: 80-band mel spectrograms
music_mels = torch.randn(1, 80, 1000)  # (batch, freq, time)
chart_mels = torch.randn(1, 80, 1000)

# Get overall quality score (0-1)
with torch.no_grad():
    score = model(music_mels, chart_mels)
    print(f"Quality Score: {score.item():.3f}")

# Get per-frame quality trace for explainability
with torch.no_grad():
    trace = model.predict_trace(music_mels, chart_mels)
    # trace shape: (batch, time)
```

## Input Specifications

- **music_mels**: `(Batch, 80, Time)` - Mel spectrogram of the music
- **chart_mels**: `(Batch, 80, Time)` - Mel spectrogram of synthesized chart audio (click sounds at note positions)

Both inputs should be normalized and have the same temporal dimensions.

## Output

- **forward()**: `(Batch,)` - Single quality score per sample in range [0, 1]
- **predict_trace()**: `(Batch, Time)` - Per-frame quality scores for interpretability

## Model Configuration

| Parameter | Default | Description |
|-----------|---------|-------------|
| `input_dim` | 80 | Mel spectrogram frequency bins |
| `d_model` | 128 | Hidden dimension |
| `n_layers` | 4 | Number of residual blocks |

## Training

The model was trained to detect misaligned or poorly-synchronized rhythm game charts by comparing music-chart pairs with various synthetic corruptions (time shifts, random note placement, etc).

## Evaluation Results

Evaluation was performed on 2,204 test samples with various segment durations. The model uses a severity parameter of 0.56.

### Overall Accuracy by Segment Duration

| Duration | Overall | Positive | Shift | Random | Mismatch |
|----------|---------|----------|-------|--------|----------|
| 5s | 81.85% | 95.69% | 79.04% | 97.41% | 97.41% |
| 10s | 83.35% | 96.55% | 80.60% | 97.41% | 100.00% |
| 20s | 84.66% | 96.55% | 82.06% | 99.14% | 100.00% |
| 30s | 85.30% | 95.69% | 82.81% | 100.00% | 100.00% |
| 60s | 85.98% | 95.69% | 83.62% | 100.00% | 100.00% |
| 120s | **86.25%** | 94.83% | **84.00%** | 100.00% | 100.00% |
| 180s | 85.57% | 94.83% | 83.19% | 100.00% | 100.00% |

### Shift Detection by Offset (120s segment)

| Offset | Accuracy | Offset | Accuracy |
|--------|----------|--------|----------|
| -0.50s | 91.38% | +0.50s | 92.24% |
| -0.30s | 89.66% | +0.30s | 89.66% |
| -0.20s | 91.38% | +0.20s | 94.83% |
| -0.10s | 100.00% | +0.10s | 100.00% |
| -0.05s | 100.00% | +0.05s | 100.00% |
| -0.03s | 91.38% | +0.03s | 95.69% |
| -0.02s | 84.48% | +0.02s | 88.79% |
| -0.01s | 20.69% | +0.01s | 13.79% |

### Analysis

The performance characteristics can be directly explained by the model's physical constraints:

1.  **Resolution Limit (±0.01s)**: Performance drops significantly here because the **10ms shift** is smaller than the model's temporal resolution (**~11.6ms per frame**). Sub-frame timing differences are mathematically difficult for the Convolutional Encoder to resolve.
2.  **Optimal Zone (±0.05s to ±0.20s)**: The model achieves **100% accuracy** here. These shifts are large enough to be resolved but small enough to fit within the **~0.36s half-receptive field**. The model can simultaneously "see" the music beat and the misaligned note, enabling a direct and precise comparison.
3.  **Field Boundary (±0.30s to ±0.50s)**: Accuracy dips slightly (to ~90%). A **0.50s shift** often pushes the note outside the receptive field of its corresponding music beat. The model can no longer compare them directly; instead, it must rely on detecting "a note without a corresponding beat" or vice-versa, which is a harder inference task (and prone to errors if the shift lands on a different valid beat).