Spaces:
Running
on
Zero
Running
on
Zero
Update gradio_app.py
Browse files- gradio_app.py +58 -46
gradio_app.py
CHANGED
|
@@ -1,4 +1,4 @@
|
|
| 1 |
-
import spaces # <---
|
| 2 |
import os
|
| 3 |
import time
|
| 4 |
import threading
|
|
@@ -9,15 +9,13 @@ import io
|
|
| 9 |
import tempfile
|
| 10 |
import numpy as np
|
| 11 |
|
| 12 |
-
#
|
| 13 |
import torch
|
| 14 |
import soundfile as sf
|
| 15 |
from pydub import AudioSegment
|
| 16 |
import gradio as gr
|
| 17 |
from fastapi import FastAPI, HTTPException
|
| 18 |
from pydantic import BaseModel
|
| 19 |
-
|
| 20 |
-
# Import thư viện nội bộ
|
| 21 |
from vieneu_tts import VieNeuTTS
|
| 22 |
|
| 23 |
# --- KHỞI TẠO FASTAPI ---
|
|
@@ -25,9 +23,11 @@ app = FastAPI()
|
|
| 25 |
|
| 26 |
print("⏳ Đang khởi động VieNeu-TTS...")
|
| 27 |
|
| 28 |
-
# --- 1. SETUP MODEL ---
|
| 29 |
-
|
| 30 |
-
|
|
|
|
|
|
|
| 31 |
|
| 32 |
# Cache
|
| 33 |
CACHE_DIR = "./reference_cache"
|
|
@@ -35,7 +35,6 @@ os.makedirs(CACHE_DIR, exist_ok=True)
|
|
| 35 |
reference_cache = {}
|
| 36 |
reference_cache_lock = threading.Lock()
|
| 37 |
|
| 38 |
-
# Hàm Cache Helper
|
| 39 |
def get_cache_path(cache_key):
|
| 40 |
key_hash = hashlib.md5(cache_key.encode()).hexdigest()
|
| 41 |
return os.path.join(CACHE_DIR, f"{key_hash}.pkl")
|
|
@@ -54,16 +53,16 @@ def save_cache_to_disk(cache_key, ref_codes):
|
|
| 54 |
with open(cache_path, 'wb') as f: pickle.dump(ref_codes, f)
|
| 55 |
except Exception: pass
|
| 56 |
|
| 57 |
-
# Load Model
|
| 58 |
try:
|
| 59 |
-
print("📦 Đang tải model vào
|
| 60 |
tts = VieNeuTTS(
|
| 61 |
backbone_repo="pnnbao-ump/VieNeu-TTS",
|
| 62 |
-
backbone_device=
|
| 63 |
codec_repo="neuphonic/neucodec",
|
| 64 |
-
codec_device=
|
| 65 |
)
|
| 66 |
-
print("✅ Model đã tải xong!")
|
| 67 |
except Exception as e:
|
| 68 |
print(f"⚠️ Lỗi tải model: {e}")
|
| 69 |
tts = None
|
|
@@ -82,12 +81,31 @@ VOICE_SAMPLES = {
|
|
| 82 |
"Nhỏ Ngọt Ngào": {"audio": "./sample/Nhỏ Ngọt Ngào.wav", "text": "./sample/Nhỏ Ngọt Ngào.txt"},
|
| 83 |
}
|
| 84 |
|
| 85 |
-
# --- 3. CORE LOGIC (
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 86 |
|
| 87 |
-
# QUAN TRỌNG: Decorator GPU
|
| 88 |
@spaces.GPU
|
| 89 |
def core_synthesize(text, voice_choice, speed_factor):
|
| 90 |
-
#
|
|
|
|
|
|
|
|
|
|
| 91 |
voice_info = VOICE_SAMPLES.get(voice_choice)
|
| 92 |
if not voice_info:
|
| 93 |
raise ValueError("Giọng không tồn tại")
|
|
@@ -95,41 +113,41 @@ def core_synthesize(text, voice_choice, speed_factor):
|
|
| 95 |
ref_audio_path = voice_info["audio"]
|
| 96 |
ref_text_path = voice_info["text"]
|
| 97 |
|
| 98 |
-
# Load reference text
|
| 99 |
with open(ref_text_path, "r", encoding="utf-8") as f:
|
| 100 |
ref_text_raw = f.read()
|
| 101 |
|
| 102 |
-
# Encode
|
| 103 |
cache_key = f"preset:{voice_choice}"
|
| 104 |
with reference_cache_lock:
|
| 105 |
if cache_key in reference_cache:
|
| 106 |
ref_codes = reference_cache[cache_key]
|
|
|
|
|
|
|
|
|
|
| 107 |
else:
|
| 108 |
ref_codes = load_cache_from_disk(cache_key)
|
| 109 |
if ref_codes is None:
|
| 110 |
-
#
|
| 111 |
-
|
| 112 |
-
|
| 113 |
-
|
| 114 |
-
|
|
|
|
|
|
|
| 115 |
reference_cache[cache_key] = ref_codes
|
| 116 |
|
| 117 |
-
# Infer
|
| 118 |
-
if torch.cuda.is_available():
|
| 119 |
-
torch.cuda.empty_cache()
|
| 120 |
wav = tts.infer(text, ref_codes, ref_text_raw)
|
| 121 |
-
|
| 122 |
-
# Speed
|
| 123 |
if speed_factor != 1.0:
|
| 124 |
with tempfile.NamedTemporaryFile(suffix='.wav', delete=False) as tmp:
|
| 125 |
sf.write(tmp.name, wav, 24000)
|
| 126 |
tmp_path = tmp.name
|
| 127 |
-
|
| 128 |
sound = AudioSegment.from_wav(tmp_path)
|
| 129 |
new_frame_rate = int(sound.frame_rate * speed_factor)
|
| 130 |
sound_stretched = sound._spawn(sound.raw_data, overrides={'frame_rate': new_frame_rate})
|
| 131 |
sound_stretched = sound_stretched.set_frame_rate(24000)
|
| 132 |
-
|
| 133 |
wav = np.array(sound_stretched.get_array_of_samples()).astype(np.float32) / 32768.0
|
| 134 |
if sound_stretched.channels == 2:
|
| 135 |
wav = wav.reshape((-1, 2)).mean(axis=1)
|
|
@@ -137,16 +155,17 @@ def core_synthesize(text, voice_choice, speed_factor):
|
|
| 137 |
|
| 138 |
return wav
|
| 139 |
|
| 140 |
-
# Hàm riêng cho Custom Voice cũng cần GPU
|
| 141 |
@spaces.GPU
|
| 142 |
def custom_synthesize_logic(text, ref_audio_path, ref_text_raw):
|
| 143 |
-
|
| 144 |
-
|
|
|
|
|
|
|
| 145 |
ref_codes = tts.encode_reference(ref_audio_path)
|
| 146 |
wav = tts.infer(text, ref_codes, ref_text_raw)
|
| 147 |
return wav
|
| 148 |
|
| 149 |
-
# --- 4. API
|
| 150 |
class FastTTSRequest(BaseModel):
|
| 151 |
text: str
|
| 152 |
voice_choice: str
|
|
@@ -161,15 +180,14 @@ async def get_voices():
|
|
| 161 |
async def fast_tts(request: FastTTSRequest):
|
| 162 |
try:
|
| 163 |
start = time.time()
|
| 164 |
-
# Gọi hàm đã
|
| 165 |
wav = core_synthesize(request.text, request.voice_choice, request.speed_factor)
|
| 166 |
process_time = time.time() - start
|
| 167 |
|
| 168 |
-
#
|
| 169 |
audio_buffer = io.BytesIO()
|
| 170 |
sf.write(audio_buffer, wav, 24000, format='WAV')
|
| 171 |
-
|
| 172 |
-
audio_base64 = base64.b64encode(audio_bytes).decode('utf-8')
|
| 173 |
|
| 174 |
return {
|
| 175 |
"status": "success",
|
|
@@ -179,7 +197,7 @@ async def fast_tts(request: FastTTSRequest):
|
|
| 179 |
except Exception as e:
|
| 180 |
raise HTTPException(status_code=500, detail=str(e))
|
| 181 |
|
| 182 |
-
# --- 5.
|
| 183 |
theme = gr.themes.Soft()
|
| 184 |
css = ".container { max-width: 900px; margin: auto; }"
|
| 185 |
|
|
@@ -190,7 +208,7 @@ def ui_synthesize(text, voice, custom_audio, custom_text, mode, speed):
|
|
| 190 |
wav = custom_synthesize_logic(text, custom_audio, custom_text)
|
| 191 |
else:
|
| 192 |
wav = core_synthesize(text, voice, speed)
|
| 193 |
-
|
| 194 |
with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp:
|
| 195 |
sf.write(tmp.name, wav, 24000)
|
| 196 |
path = tmp.name
|
|
@@ -204,17 +222,14 @@ with gr.Blocks(theme=theme, css=css, title="VieNeu-TTS") as demo:
|
|
| 204 |
with gr.Row():
|
| 205 |
with gr.Column():
|
| 206 |
inp_text = gr.Textbox(label="Văn bản", lines=3, value="Xin chào Việt Nam")
|
| 207 |
-
|
| 208 |
with gr.Tabs() as tabs:
|
| 209 |
with gr.TabItem("Giọng mẫu", id="preset_mode"):
|
| 210 |
inp_voice = gr.Dropdown(list(VOICE_SAMPLES.keys()), value="Tuyên (nam miền Bắc)", label="Chọn giọng")
|
| 211 |
with gr.TabItem("Custom", id="custom_mode"):
|
| 212 |
inp_audio = gr.Audio(type="filepath")
|
| 213 |
inp_ref_text = gr.Textbox(label="Lời thoại mẫu")
|
| 214 |
-
|
| 215 |
inp_speed = gr.Slider(0.5, 2.0, value=1.0, label="Tốc độ")
|
| 216 |
btn = gr.Button("Đọc ngay", variant="primary")
|
| 217 |
-
|
| 218 |
with gr.Column():
|
| 219 |
out_audio = gr.Audio(label="Kết quả", autoplay=True)
|
| 220 |
out_status = gr.Textbox(label="Trạng thái")
|
|
@@ -222,13 +237,10 @@ with gr.Blocks(theme=theme, css=css, title="VieNeu-TTS") as demo:
|
|
| 222 |
mode_state = gr.Textbox(visible=False, value="preset_mode")
|
| 223 |
tabs.children[0].select(lambda: "preset_mode", None, mode_state)
|
| 224 |
tabs.children[1].select(lambda: "custom_mode", None, mode_state)
|
| 225 |
-
|
| 226 |
btn.click(ui_synthesize, [inp_text, inp_voice, inp_audio, inp_ref_text, mode_state, inp_speed], [out_audio, out_status])
|
| 227 |
|
| 228 |
-
# --- 6. MOUNT GRADIO VÀO FASTAPI ---
|
| 229 |
app = gr.mount_gradio_app(app, demo, path="/")
|
| 230 |
|
| 231 |
-
# --- 7. CHẠY SERVER ---
|
| 232 |
if __name__ == "__main__":
|
| 233 |
import uvicorn
|
| 234 |
uvicorn.run(app, host="0.0.0.0", port=7860)
|
|
|
|
| 1 |
+
import spaces # <--- LUÔN ĐỂ ĐẦU TIÊN
|
| 2 |
import os
|
| 3 |
import time
|
| 4 |
import threading
|
|
|
|
| 9 |
import tempfile
|
| 10 |
import numpy as np
|
| 11 |
|
| 12 |
+
# Import các thư viện khác
|
| 13 |
import torch
|
| 14 |
import soundfile as sf
|
| 15 |
from pydub import AudioSegment
|
| 16 |
import gradio as gr
|
| 17 |
from fastapi import FastAPI, HTTPException
|
| 18 |
from pydantic import BaseModel
|
|
|
|
|
|
|
| 19 |
from vieneu_tts import VieNeuTTS
|
| 20 |
|
| 21 |
# --- KHỞI TẠO FASTAPI ---
|
|
|
|
| 23 |
|
| 24 |
print("⏳ Đang khởi động VieNeu-TTS...")
|
| 25 |
|
| 26 |
+
# --- 1. SETUP MODEL (SỬA LẠI CHO ZEROGPU) ---
|
| 27 |
+
# QUAN TRỌNG: Trên ZeroGPU, lúc khởi động PHẢI DÙNG CPU
|
| 28 |
+
# GPU chỉ được kích hoạt bên trong hàm @spaces.GPU
|
| 29 |
+
device = "cpu"
|
| 30 |
+
print(f"🖥️ Thiết bị khởi động (Global): {device.upper()} (Sẽ chuyển sang CUDA khi chạy)")
|
| 31 |
|
| 32 |
# Cache
|
| 33 |
CACHE_DIR = "./reference_cache"
|
|
|
|
| 35 |
reference_cache = {}
|
| 36 |
reference_cache_lock = threading.Lock()
|
| 37 |
|
|
|
|
| 38 |
def get_cache_path(cache_key):
|
| 39 |
key_hash = hashlib.md5(cache_key.encode()).hexdigest()
|
| 40 |
return os.path.join(CACHE_DIR, f"{key_hash}.pkl")
|
|
|
|
| 53 |
with open(cache_path, 'wb') as f: pickle.dump(ref_codes, f)
|
| 54 |
except Exception: pass
|
| 55 |
|
| 56 |
+
# Load Model vào CPU trước
|
| 57 |
try:
|
| 58 |
+
print("📦 Đang tải model vào RAM (CPU)...")
|
| 59 |
tts = VieNeuTTS(
|
| 60 |
backbone_repo="pnnbao-ump/VieNeu-TTS",
|
| 61 |
+
backbone_device="cpu", # Bắt buộc là CPU
|
| 62 |
codec_repo="neuphonic/neucodec",
|
| 63 |
+
codec_device="cpu" # Bắt buộc là CPU
|
| 64 |
)
|
| 65 |
+
print("✅ Model đã tải xong (Ready on CPU)!")
|
| 66 |
except Exception as e:
|
| 67 |
print(f"⚠️ Lỗi tải model: {e}")
|
| 68 |
tts = None
|
|
|
|
| 81 |
"Nhỏ Ngọt Ngào": {"audio": "./sample/Nhỏ Ngọt Ngào.wav", "text": "./sample/Nhỏ Ngọt Ngào.txt"},
|
| 82 |
}
|
| 83 |
|
| 84 |
+
# --- 3. CORE LOGIC (ZeroGPU Optimization) ---
|
| 85 |
+
|
| 86 |
+
def move_model_to_cuda():
|
| 87 |
+
"""Hàm helper để đẩy model sang GPU khi cần"""
|
| 88 |
+
if torch.cuda.is_available():
|
| 89 |
+
# Kiểm tra xem model đã ở trên GPU chưa để tránh move thừa
|
| 90 |
+
# VieNeuTTS lưu model trong self.backbone và self.codec
|
| 91 |
+
try:
|
| 92 |
+
# Move backbone
|
| 93 |
+
if next(tts.backbone.parameters()).device.type != 'cuda':
|
| 94 |
+
print(" 🚀 Moving model to GPU...")
|
| 95 |
+
tts.backbone.to("cuda")
|
| 96 |
+
|
| 97 |
+
# Move codec
|
| 98 |
+
if next(tts.codec.parameters()).device.type != 'cuda':
|
| 99 |
+
tts.codec.to("cuda")
|
| 100 |
+
except Exception as e:
|
| 101 |
+
print(f"⚠️ Lỗi khi move model sang GPU: {e}")
|
| 102 |
|
|
|
|
| 103 |
@spaces.GPU
|
| 104 |
def core_synthesize(text, voice_choice, speed_factor):
|
| 105 |
+
# 1. Đẩy model sang GPU (Chỉ làm việc này bên trong hàm @spaces.GPU)
|
| 106 |
+
move_model_to_cuda()
|
| 107 |
+
|
| 108 |
+
# 2. Lấy thông tin giọng
|
| 109 |
voice_info = VOICE_SAMPLES.get(voice_choice)
|
| 110 |
if not voice_info:
|
| 111 |
raise ValueError("Giọng không tồn tại")
|
|
|
|
| 113 |
ref_audio_path = voice_info["audio"]
|
| 114 |
ref_text_path = voice_info["text"]
|
| 115 |
|
|
|
|
| 116 |
with open(ref_text_path, "r", encoding="utf-8") as f:
|
| 117 |
ref_text_raw = f.read()
|
| 118 |
|
| 119 |
+
# 3. Encode Reference
|
| 120 |
cache_key = f"preset:{voice_choice}"
|
| 121 |
with reference_cache_lock:
|
| 122 |
if cache_key in reference_cache:
|
| 123 |
ref_codes = reference_cache[cache_key]
|
| 124 |
+
# Đảm bảo ref_codes cũng ở trên GPU
|
| 125 |
+
if isinstance(ref_codes, torch.Tensor):
|
| 126 |
+
ref_codes = ref_codes.to("cuda")
|
| 127 |
else:
|
| 128 |
ref_codes = load_cache_from_disk(cache_key)
|
| 129 |
if ref_codes is None:
|
| 130 |
+
ref_codes = tts.encode_reference(ref_audio_path) # Lúc này model đã ở GPU nên encode sẽ nhanh
|
| 131 |
+
# Move về CPU để cache
|
| 132 |
+
save_cache_to_disk(cache_key, ref_codes.cpu() if isinstance(ref_codes, torch.Tensor) else ref_codes)
|
| 133 |
+
|
| 134 |
+
# Đẩy lại lên GPU để dùng
|
| 135 |
+
if isinstance(ref_codes, torch.Tensor):
|
| 136 |
+
ref_codes = ref_codes.to("cuda")
|
| 137 |
reference_cache[cache_key] = ref_codes
|
| 138 |
|
| 139 |
+
# 4. Infer
|
|
|
|
|
|
|
| 140 |
wav = tts.infer(text, ref_codes, ref_text_raw)
|
| 141 |
+
|
| 142 |
+
# 5. Speed Control (CPU)
|
| 143 |
if speed_factor != 1.0:
|
| 144 |
with tempfile.NamedTemporaryFile(suffix='.wav', delete=False) as tmp:
|
| 145 |
sf.write(tmp.name, wav, 24000)
|
| 146 |
tmp_path = tmp.name
|
|
|
|
| 147 |
sound = AudioSegment.from_wav(tmp_path)
|
| 148 |
new_frame_rate = int(sound.frame_rate * speed_factor)
|
| 149 |
sound_stretched = sound._spawn(sound.raw_data, overrides={'frame_rate': new_frame_rate})
|
| 150 |
sound_stretched = sound_stretched.set_frame_rate(24000)
|
|
|
|
| 151 |
wav = np.array(sound_stretched.get_array_of_samples()).astype(np.float32) / 32768.0
|
| 152 |
if sound_stretched.channels == 2:
|
| 153 |
wav = wav.reshape((-1, 2)).mean(axis=1)
|
|
|
|
| 155 |
|
| 156 |
return wav
|
| 157 |
|
|
|
|
| 158 |
@spaces.GPU
|
| 159 |
def custom_synthesize_logic(text, ref_audio_path, ref_text_raw):
|
| 160 |
+
# 1. Đẩy model sang GPU
|
| 161 |
+
move_model_to_cuda()
|
| 162 |
+
|
| 163 |
+
# 2. Xử lý
|
| 164 |
ref_codes = tts.encode_reference(ref_audio_path)
|
| 165 |
wav = tts.infer(text, ref_codes, ref_text_raw)
|
| 166 |
return wav
|
| 167 |
|
| 168 |
+
# --- 4. API ---
|
| 169 |
class FastTTSRequest(BaseModel):
|
| 170 |
text: str
|
| 171 |
voice_choice: str
|
|
|
|
| 180 |
async def fast_tts(request: FastTTSRequest):
|
| 181 |
try:
|
| 182 |
start = time.time()
|
| 183 |
+
# Gọi hàm đã decorate
|
| 184 |
wav = core_synthesize(request.text, request.voice_choice, request.speed_factor)
|
| 185 |
process_time = time.time() - start
|
| 186 |
|
| 187 |
+
# Base64
|
| 188 |
audio_buffer = io.BytesIO()
|
| 189 |
sf.write(audio_buffer, wav, 24000, format='WAV')
|
| 190 |
+
audio_base64 = base64.b64encode(audio_buffer.getvalue()).decode('utf-8')
|
|
|
|
| 191 |
|
| 192 |
return {
|
| 193 |
"status": "success",
|
|
|
|
| 197 |
except Exception as e:
|
| 198 |
raise HTTPException(status_code=500, detail=str(e))
|
| 199 |
|
| 200 |
+
# --- 5. UI ---
|
| 201 |
theme = gr.themes.Soft()
|
| 202 |
css = ".container { max-width: 900px; margin: auto; }"
|
| 203 |
|
|
|
|
| 208 |
wav = custom_synthesize_logic(text, custom_audio, custom_text)
|
| 209 |
else:
|
| 210 |
wav = core_synthesize(text, voice, speed)
|
| 211 |
+
|
| 212 |
with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp:
|
| 213 |
sf.write(tmp.name, wav, 24000)
|
| 214 |
path = tmp.name
|
|
|
|
| 222 |
with gr.Row():
|
| 223 |
with gr.Column():
|
| 224 |
inp_text = gr.Textbox(label="Văn bản", lines=3, value="Xin chào Việt Nam")
|
|
|
|
| 225 |
with gr.Tabs() as tabs:
|
| 226 |
with gr.TabItem("Giọng mẫu", id="preset_mode"):
|
| 227 |
inp_voice = gr.Dropdown(list(VOICE_SAMPLES.keys()), value="Tuyên (nam miền Bắc)", label="Chọn giọng")
|
| 228 |
with gr.TabItem("Custom", id="custom_mode"):
|
| 229 |
inp_audio = gr.Audio(type="filepath")
|
| 230 |
inp_ref_text = gr.Textbox(label="Lời thoại mẫu")
|
|
|
|
| 231 |
inp_speed = gr.Slider(0.5, 2.0, value=1.0, label="Tốc độ")
|
| 232 |
btn = gr.Button("Đọc ngay", variant="primary")
|
|
|
|
| 233 |
with gr.Column():
|
| 234 |
out_audio = gr.Audio(label="Kết quả", autoplay=True)
|
| 235 |
out_status = gr.Textbox(label="Trạng thái")
|
|
|
|
| 237 |
mode_state = gr.Textbox(visible=False, value="preset_mode")
|
| 238 |
tabs.children[0].select(lambda: "preset_mode", None, mode_state)
|
| 239 |
tabs.children[1].select(lambda: "custom_mode", None, mode_state)
|
|
|
|
| 240 |
btn.click(ui_synthesize, [inp_text, inp_voice, inp_audio, inp_ref_text, mode_state, inp_speed], [out_audio, out_status])
|
| 241 |
|
|
|
|
| 242 |
app = gr.mount_gradio_app(app, demo, path="/")
|
| 243 |
|
|
|
|
| 244 |
if __name__ == "__main__":
|
| 245 |
import uvicorn
|
| 246 |
uvicorn.run(app, host="0.0.0.0", port=7860)
|