metricsresult.py
1.22 KB
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
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
# Charger le fichier CSV
df = pd.read_csv("audio_comparison_results_normalized.csv")
# Calculer les moyennes pour les différentes métriques
mse_vocals_target_avg = df["mse_vocals_target"].mean()
snr_vocals_target_avg = df["snr_vocals_target"].mean()
sdr_vocals_target_avg = df["sdr_vocals_target"].mean()
mse_mixture_target_avg = df["mse_mixture_target"].mean()
snr_mixture_target_avg = df["snr_mixture_target"].mean()
sdr_mixture_target_avg = df["sdr_mixture_target"].mean()
# Construire le tableau des valeurs
metrics_table = [
["DemucsV3", "SNR", "SDR", "MSE"],
["vocals vs target", f"{snr_vocals_target_avg:.2f}", f"{sdr_vocals_target_avg:.2f}", f"{mse_vocals_target_avg:.2f}"],
["mixture vs target", f"{snr_mixture_target_avg:.2f}", f"{sdr_mixture_target_avg:.2f}", f"{mse_mixture_target_avg:.2f}"]
]
# Créer la figure et l'axe pour le tableau
fig, ax = plt.subplots(figsize=(8, 3))
ax.axis('off') # Masquer les axes
# Créer le tableau
ax.table(
cellText=metrics_table,
colLabels=None,
cellLoc='center',
loc='center',
colWidths=[0.2] * 4
)
# Sauvegarder l'image du tableau
plt.savefig("metrics.jpg", bbox_inches='tight')
plt.show()