config.py

"""
Configuration for the listening test.
Update SAMPLES with your actual audio paths and instructions.
Model names map to the keys used in results JSON.
"""

MODEL_NAMES = {
    "target": "Target",
    "model_a": "Baseline",
    "model_b": "Finetuned",
    "model_c": "SAO-Instruct",
}

INSTRUCTIONS = [
	"add bossa nova electric bass",
	"include full acoustic drumkit",
	"insert clean electric guitar",
	"layer piano, organ",
	"Include reggae Saxophone",
	"remove bass guitar",
	"delete tomtom, cymbal, hihat, snare, kick",
	"mute clean electric guitar, guitars",
	"minus grand piano, electric piano",
	"Omit blues Lead male vocal"
]

SAMPLES = [
    {
        "id": f"sample_{i:02d}",
        "instruction": INSTRUCTIONS[i - 1],
        "input_audio": f"audio/sample_{i:02d}/input.wav",
        "target_audio": f"audio/sample_{i:02d}/target.wav",
        "model_a_audio": f"audio/sample_{i:02d}/model_a.wav",
        "model_b_audio": f"audio/sample_{i:02d}/model_b.wav",
        "model_c_audio": f"audio/sample_{i:02d}/model_c.wav",
    }
    for i in range(1, 11)
]

CLIP_KEYS = ["target", "model_a", "model_b", "model_c"]

RATING_CATEGORIES = {
    "quality": "Quality - Quality of the edited audio.",
    "relevance": "Relevance - How well the edit matches the instruction.",
    "faithfulness": "Faithfulness - How well unedited parts of the audio are preserved.",
}

MOS_SCALE = {
    1: "1 Bad",
    2: "2 Poor",
    3: "3 Fair",
    4: "4 Good",
    5: "5 Excellent",
}