File size: 10,120 Bytes
97bd888
222cbc4
97bd888
222cbc4
 
 
 
 
792e819
222cbc4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
97bd888
222cbc4
97bd888
222cbc4
 
792e819
222cbc4
 
fbf325c
222cbc4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
97bd888
 
 
 
222cbc4
 
 
792e819
222cbc4
 
97bd888
792e819
 
 
 
 
 
 
 
 
 
 
 
222cbc4
 
97bd888
222cbc4
 
97bd888
222cbc4
97bd888
222cbc4
 
792e819
222cbc4
 
792e819
 
 
 
 
 
 
97bd888
222cbc4
 
 
 
 
792e819
222cbc4
97bd888
222cbc4
792e819
97bd888
222cbc4
792e819
222cbc4
97bd888
222cbc4
792e819
1fd11d5
97bd888
 
 
792e819
222cbc4
 
 
792e819
1fd11d5
792e819
 
 
1fd11d5
 
 
792e819
 
 
222cbc4
792e819
 
222cbc4
 
1fd11d5
792e819
222cbc4
97bd888
 
792e819
97bd888
 
 
792e819
 
 
97bd888
222cbc4
792e819
97bd888
792e819
97bd888
1fd11d5
792e819
97bd888
 
 
792e819
97bd888
 
1fd11d5
97bd888
 
792e819
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
97bd888
 
 
 
792e819
97bd888
1fd11d5
792e819
 
97bd888
792e819
97bd888
1fd11d5
222cbc4
 
792e819
 
222cbc4
 
 
 
 
792e819
222cbc4
 
 
 
97bd888
 
222cbc4
 
 
792e819
222cbc4
 
 
 
792e819
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
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
import spaces 
import os
os.environ['SPACES_ZERO_GPU'] = '1'

import gradio as gr
import soundfile as sf
import tempfile
import torch
import librosa # Thêm thư viện xử lý âm thanh
from vieneu_tts import VieNeuTTS
import time

# --- 1. SETUP MODEL ---
device = "cuda" if torch.cuda.is_available() else "cpu"

try:
    tts = VieNeuTTS(
        backbone_repo="pnnbao-ump/VieNeu-TTS",
        backbone_device=device,
        codec_repo="neuphonic/neucodec",
        codec_device=device
    )
except Exception as e:
    class MockTTS:
        def encode_reference(self, path): return None
        def infer(self, text, ref, ref_text): 
            time.sleep(1.2) 
            import numpy as np
            return np.random.uniform(-0.1, 0.1, 24000*2)
    tts = MockTTS()

# --- 2. DATA (Giữ nguyên danh sách giọng mẫu) ---
VOICE_SAMPLES = {
    "Tuyên (nam miền Bắc)": {"audio": "./sample/Tuyên (nam miền Bắc).wav", "text": "./sample/Tuyên (nam miền Bắc).txt"},
    "Thiện Tâm": {"audio": "./sample/thientam.mp3", "text": "./sample/thientam.txt"},
    "Vĩnh (nam miền Nam)": {"audio": "./sample/Vĩnh (nam miền Nam).wav", "text": "./sample/Vĩnh (nam miền Nam).txt"},
    "Bình (nam miền Bắc)": {"audio": "./sample/Bình (nam miền Bắc).wav", "text": "./sample/Bình (nam miền Bắc).txt"},
    "Nguyên (nam miền Nam)": {"audio": "./sample/Nguyên (nam miền Nam).wav", "text": "./sample/Nguyên (nam miền Nam).txt"},
    "Sơn (nam miền Nam)": {"audio": "./sample/Sơn (nam miền Nam).wav", "text": "./sample/Sơn (nam miền Nam).txt"},
    "Đoan (nữ miền Nam)": {"audio": "./sample/Đoan (nữ miền Nam).wav", "text": "./sample/Đoan (nữ miền Nam).txt"},
    "Ngọc (nữ miền Bắc)": {"audio": "./sample/Ngọc (nữ miền Bắc).wav", "text": "./sample/Ngọc (nữ miền Bắc).txt"},
    "Ly (nữ miền Bắc)": {"audio": "./sample/Ly (nữ miền Bắc).wav", "text": "./sample/Ly (nữ miền Bắc).txt"},
    "Dung (nữ miền Nam)": {"audio": "./sample/Dung (nữ miền Nam).wav", "text": "./sample/Dung (nữ miền Nam).txt"}
}

# --- 3. HELPER FUNCTIONS ---
def load_reference_info(voice_choice):
    if voice_choice in VOICE_SAMPLES:
        audio_path = VOICE_SAMPLES[voice_choice]["audio"]
        text_path = VOICE_SAMPLES[voice_choice]["text"]
        if os.path.exists(text_path):
            with open(text_path, "r", encoding="utf-8") as f:
                ref_text = f.read()
            return audio_path, ref_text
    return None, ""

@spaces.GPU(duration=120) 
def synthesize_speech(text, voice_choice, custom_audio, custom_text, mode_tab, pause_level, speed_value):
    try:
        if not text or text.strip() == "":
            return None, "⚠️ Vui lòng nhập nội dung!"
        
        # 3.1. Xử lý độ ngắt nghỉ (Pause level)
        processed_text = text
        if pause_level == "Trung bình":
            processed_text = processed_text.replace(",", ", , ").replace(".", ". . ")
        elif pause_level == "Dài":
            processed_text = processed_text.replace(",", ", , , ").replace(".", ". . . . ")
            
        if len(processed_text) > 400:
             processed_text = processed_text[:400]

        # 3.2. Lấy dữ liệu Reference
        if mode_tab == "custom_mode": 
            if custom_audio is None or not custom_text:
                return None, "⚠️ Thiếu Audio mẫu hoặc Text mẫu."
            ref_audio_path = custom_audio
            ref_text_raw = custom_text
        else:
            ref_audio_path = VOICE_SAMPLES[voice_choice]["audio"]
            with open(VOICE_SAMPLES[voice_choice]["text"], "r", encoding="utf-8") as f:
                ref_text_raw = f.read()

        # 3.3. Thực hiện Inference
        start_time = time.time()
        ref_codes = tts.encode_reference(ref_audio_path)
        wav = tts.infer(processed_text, ref_codes, ref_text_raw)
        
        # 3.4. Điều chỉnh Tốc độ (Speed) bằng librosa
        if speed_value != 1.0:
            # Time stretch giữ nguyên pitch
            wav = librosa.effects.time_stretch(wav, rate=float(speed_value))
            
        process_time = time.time() - start_time
        
        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
            sf.write(tmp_file.name, wav, 24000)
            output_path = tmp_file.name
        
        return output_path, f"⚡ Xử lý: {process_time:.2f}s | Tốc độ: {speed_value}x"
    except Exception as e:
        return None, f"❌ Lỗi: {str(e)}"

# --- 4. THEME & CSS ---
theme = gr.themes.Default(
    primary_hue="indigo",
    secondary_hue="blue",
    neutral_hue="slate",
    font=[gr.themes.GoogleFont('Inter'), 'sans-serif'],
).set(
    body_background_fill="#020617",
    block_background_fill="#0f172a",
    block_border_width="1px",
    input_background_fill="#1e293b",
    input_border_color="#334155",
    button_primary_background_fill="linear-gradient(135deg, #4f46e5 0%, #7c3aed 100%)",
)

css = """
.main-wrap { max-width: 1200px !important; margin: auto !important; padding: 20px !important; }
.st-card { 
    border-radius: 16px !important; 
    border: 1px solid rgba(255,255,255,0.1) !important; 
    box-shadow: 0 4px 20px rgba(0,0,0,0.5) !important;
    padding: 15px;
}
.result-card {
    background: linear-gradient(180deg, rgba(15, 23, 42, 0.8) 0%, rgba(30, 41, 59, 0.8) 100%) !important;
    border: 1px solid rgba(99, 102, 241, 0.2) !important;
    margin-top: 15px;
}
audio { filter: invert(90%) hue-rotate(180deg) brightness(1.5); width: 100%; border-radius: 8px; }
.footer { text-align: center; margin-top: 40px; color: #475569; font-size: 0.8rem; font-weight: 500; }
"""

# --- 5. UI CONSTRUCTION ---
with gr.Blocks(title="AI Voice Studio") as demo:
    
    with gr.Column(elem_classes="main-wrap"):
        with gr.Row(equal_height=True):
            # CỘT TRÁI
            with gr.Column(scale=1):
                with gr.Group(elem_classes="st-card"):
                    text_input = gr.Textbox(
                        label="VĂN BẢN CẦN CHUYỂN ĐỔI",
                        placeholder="Nhập nội dung vào đây...",
                        lines=20, # Tăng thêm để cân bằng với các nút mới
                        show_label=True,
                    )
                    char_count = gr.HTML("<div style='text-align: right; color: #6366f1; font-size: 0.85rem; font-weight: bold; padding: 5px;'>0 / 250</div>")

            # CỘT PHẢI
            with gr.Column(scale=1):
                with gr.Tabs() as tabs:
                    with gr.TabItem("👤 Nghệ sĩ đọc", id="preset_mode"):
                        voice_select = gr.Dropdown(
                            choices=list(VOICE_SAMPLES.keys()),
                            value="Tuyên (nam miền Bắc)",
                            label="Lựa chọn giọng đọc mẫu",
                        )
                        with gr.Accordion("Nghe thử giọng mẫu", open=False):
                            ref_audio_preview = gr.Audio(interactive=False, show_label=False)
                            ref_text_preview = gr.Markdown("...")

                    with gr.TabItem("🎙️ Nhân bản (Clone)", id="custom_mode"):
                        custom_audio = gr.Audio(label="Audio gốc", type="filepath")
                        custom_text = gr.Textbox(
                            label="NỘI DUNG AUDIO MẪU", 
                            placeholder="Nhập lời thoại của audio mẫu...",
                            lines=4, 
                            show_label=True
                        )

                # --- KHU VỰC ĐIỀU CHỈNH ÂM THANH ---
                with gr.Row():
                    pause_level = gr.Radio(
                        choices=["Mặc định", "Trung bình", "Dài"], 
                        value="Mặc định", 
                        label="Độ ngắt nghỉ",
                        scale=1
                    )
                    speed_select = gr.Dropdown(
                        choices=[0.8, 0.9, 1.0, 1.1, 1.2, 1.5],
                        value=1.0,
                        label="Tốc độ đọc",
                        scale=1
                    )

                current_mode = gr.State(value="preset_mode")
                
                gr.Markdown("<br>")
                btn_generate = gr.Button("BẮT ĐẦU TỔNG HỢP", variant="primary", size="lg")
                
                with gr.Group(elem_classes="st-card result-card"):
                    audio_output = gr.Audio(label="AUDIO KẾT QUẢ", interactive=False, autoplay=True)
                    status_output = gr.Markdown("<p style='text-align: center; color: #818cf8; font-weight: 500;'>✨ Sẵn sàng thực hiện</p>")

        gr.HTML("<div class='footer'>ENGINE BY VIENEU-TTS • PROFESSIONAL AI SOLUTIONS 2025</div>")

    # --- LOGIC ---
    def update_count(text):
        l = len(text)
        color = "#6366f1" if l <= 250 else "#f43f5e"
        return f"<div style='text-align: right; color: {color}; font-size: 0.85rem; font-weight: bold; padding: 5px;'>{l} / 250</div>"

    text_input.change(update_count, text_input, char_count)

    def update_ref_preview(voice):
        audio, text = load_reference_info(voice)
        return audio, f"**Nội dung mẫu:** *\"{text}\"*"
    
    voice_select.change(update_ref_preview, voice_select, [ref_audio_preview, ref_text_preview])
    demo.load(update_ref_preview, voice_select, [ref_audio_preview, ref_text_preview])

    tabs.children[0].select(fn=lambda: "preset_mode", outputs=current_mode)
    tabs.children[1].select(fn=lambda: "custom_mode", outputs=current_mode)

    btn_generate.click(
        fn=synthesize_speech,
        inputs=[text_input, voice_select, custom_audio, custom_text, current_mode, pause_level, speed_select],
        outputs=[audio_output, status_output]
    )

if __name__ == "__main__":
    demo.queue().launch(theme=theme, css=css, server_name="0.0.0.0", server_port=7860)