File size: 5,038 Bytes
b2d0c3c
 
 
 
 
 
 
 
 
 
 
 
6eb335e
b2d0c3c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6eb335e
b2d0c3c
 
 
 
 
6eb335e
b2d0c3c
 
6eb335e
b2d0c3c
 
 
 
 
 
6eb335e
b2d0c3c
 
6eb335e
b2d0c3c
 
 
 
 
 
 
 
6eb335e
b2d0c3c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6eb335e
b2d0c3c
 
6eb335e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b2d0c3c
6eb335e
 
 
 
 
 
 
b2d0c3c
 
6eb335e
b2d0c3c
 
6eb335e
 
 
 
b2d0c3c
6eb335e
b2d0c3c
6eb335e
 
 
b2d0c3c
241a38a
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
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
import gradio as gr
import torch
import torch.nn as nn
import soundfile as sf
import numpy as np
import torchaudio
import os
import gc
import time
from demucs.apply import apply_model
from demucs.pretrained import get_model

print("🚀 Starting SpecTacles...")

# Force CPU for free hosting
device = "cpu"

# ==========================================
# 1. DEFINE BRAIN
# ==========================================
class StemMixer(nn.Module):
    def __init__(self):
        super().__init__()
        self.n_fft, self.hop = 2048, 512
        self.mixer = nn.Sequential(
            nn.Linear(4, 64), nn.BatchNorm1d(64), nn.ReLU(),
            nn.Linear(64, 32), nn.BatchNorm1d(32), nn.ReLU(),
            nn.Linear(32, 16), nn.BatchNorm1d(16), nn.ReLU(),
            nn.Linear(16, 2), nn.Sigmoid()
        )
    def compute_spec(self, x):
        b, s, c, t = x.shape
        x = x.reshape(b * s * c, t)
        return torch.abs(torch.stft(x, self.n_fft, self.hop, window=torch.hann_window(self.n_fft).to(x.device), return_complex=True))
    def forward(self, d, m):
        ds, ms = self.compute_spec(d), self.compute_spec(m)
        b, s, c, t = d.shape
        f, fr = ds.shape[1], ds.shape[2]
        ds, ms = ds.reshape(b, s, c, f, fr), ms.reshape(b, s, c, f, fr)
        inp = torch.cat([ds, ms], dim=2).permute(0, 1, 3, 4, 2).reshape(-1, 4)
        mask = self.mixer(inp).reshape(b, s, f, fr, c).permute(0, 1, 4, 2, 3)
        return mask * ds + (1 - mask) * ms

# ==========================================
# 2. LIGHTWEIGHT STARTUP
# ==========================================
print("Loading Brain...")
mixer = StemMixer().to(device)

if os.path.exists("best_stem_mixer.pth"):
    mixer.load_state_dict(torch.load("best_stem_mixer.pth", map_location=torch.device('cpu')))
else:
    print("⚠️ Model file missing!")
mixer.eval()

# Lazy Loading Variables
model_d = None
model_m = None

def load_heavy_models():
    global model_d, model_m
    if model_d is None:
        print("⏳ Loading Demucs...")
        model_d = get_model('htdemucs').to(device).eval()
    if model_m is None:
        print("⏳ Loading MDX...")
        model_m = get_model('mdx_extra').to(device).eval()

# ==========================================
# 3. PROCESS FUNCTION
# ==========================================
def process(audio):
    if audio is None: return None
    load_heavy_models()
    print(f"Processing...")
    
    wav, sr = torchaudio.load(audio)
    if sr != 44100: wav = torchaudio.transforms.Resample(sr, 44100)(wav)
    if wav.abs().max() > 1.0: wav = wav / wav.abs().max()
    
    chunk = 44100 * 10 
    stems = []
    
    for s in range(0, wav.shape[1], chunk):
        e = min(s + chunk, wav.shape[1])
        c = wav[:, s:e]
        if c.shape[1] < 2048: continue
        
        with torch.no_grad():
            ct = c.unsqueeze(0).to(device)
            d = apply_model(model_d, ct, shifts=0)
            m = apply_model(model_m, ct, shifts=0)
            mix = mixer(d, m)
            
            b,st,ch,t = d.shape
            dc = torch.stft(d.reshape(b*st*ch, t), 2048, 512, window=torch.hann_window(2048).to(device), return_complex=True)
            rec = torch.istft(torch.polar(mix.reshape(b*st*ch, dc.shape[-2], dc.shape[-1]), torch.angle(dc)), 2048, 512, window=torch.hann_window(2048).to(device), length=t)
            stems.append(rec.reshape(st, ch, -1).cpu())

    full = torch.cat(stems, dim=2).numpy()
    paths = []
    ts = int(time.time())
    
    for i, n in enumerate(['Drums', 'Bass', 'Other', 'Vocals']):
        p = f"{n}_{ts}.mp3"
        sf.write(p, np.clip(full[i].T, -0.99, 0.99), 44100)
        paths.append(p)
        
    return paths[3], paths[0], paths[1], paths[2]

# ==========================================
# 4. UI (Nonchalant & Themed)
# ==========================================

# JavaScript to toggle Dark Mode
js_toggle = """
function() {
    document.body.classList.toggle('dark');
}
"""

# CSS for the button and clean look
css = """
.gr-button-primary {
    background: linear-gradient(90deg, #6366f1, #a855f7) !important; 
    color: white !important; 
    border: none !important;
}
h1 {
    font-family: 'Helvetica', sans-serif; 
    font-weight: 700;
}
"""

# Use Soft theme (Looks good in both Light and Dark)
with gr.Blocks(theme=gr.themes.Soft(), css=css) as app:
    
    with gr.Row():
        with gr.Column(scale=10):
            gr.Markdown("# 👓 SpecTacles")
        with gr.Column(scale=1):
            # The Theme Toggle Button
            theme_btn = gr.Button("🌗", variant="secondary")

    with gr.Row():
        inp = gr.Audio(type="filepath", label="Source")
        btn = gr.Button("SEPARATE", variant="primary")
    
    with gr.Row():
        v = gr.Audio(label="Vocals")
        d = gr.Audio(label="Drums")
        b = gr.Audio(label="Bass")
        o = gr.Audio(label="Other")
        
    # Logic
    btn.click(process, inputs=inp, outputs=[v,d,b,o])
    
    # JS Trigger
    theme_btn.click(None, None, None, js=js_toggle)

app.launch()