Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -6,6 +6,7 @@ import tempfile
|
|
| 6 |
import shutil
|
| 7 |
import asyncio
|
| 8 |
import requests
|
|
|
|
| 9 |
from pydub import AudioSegment
|
| 10 |
import yt_dlp
|
| 11 |
from google import genai
|
|
@@ -61,47 +62,53 @@ def download_youtube_video(url, output_path):
|
|
| 61 |
return output_path
|
| 62 |
|
| 63 |
def extract_audio_from_video(video_path, audio_path):
|
| 64 |
-
|
|
|
|
| 65 |
return audio_path
|
| 66 |
|
| 67 |
def get_video_duration(video_path):
|
| 68 |
result = subprocess.run(['ffprobe', '-v', 'error', '-show_entries', 'format=duration', '-of', 'default=noprint_wrappers=1:nokey=1', video_path], capture_output=True, text=True, check=True)
|
| 69 |
return float(result.stdout.strip())
|
| 70 |
|
| 71 |
-
def remove_silence_from_edges(audio_segment, silence_thresh=-45.0, chunk_size=10):
|
| 72 |
-
if len(audio_segment) < 100: return audio_segment
|
| 73 |
-
trim_ms = 0
|
| 74 |
-
while trim_ms < len(audio_segment) and audio_segment[trim_ms:trim_ms+chunk_size].dBFS < silence_thresh: trim_ms += chunk_size
|
| 75 |
-
start_trim = trim_ms
|
| 76 |
-
trim_ms = 0
|
| 77 |
-
while trim_ms < len(audio_segment) and audio_segment[len(audio_segment)-trim_ms-chunk_size:len(audio_segment)-trim_ms].dBFS < silence_thresh: trim_ms += chunk_size
|
| 78 |
-
end_trim = len(audio_segment) - trim_ms
|
| 79 |
-
return audio_segment[start_trim:end_trim]
|
| 80 |
-
|
| 81 |
def adjust_audio_speed_ffmpeg(input_wav, output_wav, target_duration):
|
| 82 |
try:
|
| 83 |
audio = AudioSegment.from_file(input_wav)
|
| 84 |
-
|
| 85 |
-
if len(
|
| 86 |
-
|
| 87 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 88 |
|
| 89 |
-
|
|
|
|
|
|
|
|
|
|
| 90 |
if orig_dur <= 0.1 or target_duration <= 0.1:
|
| 91 |
shutil.copy(temp_trimmed, output_wav); return
|
| 92 |
-
|
| 93 |
speed_factor = orig_dur / target_duration
|
|
|
|
| 94 |
if speed_factor > 2.0: speed_factor = 2.0
|
| 95 |
if speed_factor < 0.6: speed_factor = 0.6
|
| 96 |
-
|
| 97 |
atempo_filters = []
|
| 98 |
current = speed_factor
|
| 99 |
while current > 2.0: atempo_filters.append("atempo=2.0"); current /= 2.0
|
| 100 |
while current < 0.5: atempo_filters.append("atempo=0.5"); current /= 0.5
|
| 101 |
if current != 1.0: atempo_filters.append(f"atempo={current}")
|
| 102 |
-
|
| 103 |
if not atempo_filters: shutil.copy(temp_trimmed, output_wav); return
|
| 104 |
-
|
| 105 |
subprocess.run(['ffmpeg', '-y', '-i', temp_trimmed, '-filter:a', ",".join(atempo_filters), output_wav], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, check=True)
|
| 106 |
except: shutil.copy(input_wav, output_wav)
|
| 107 |
|
|
@@ -132,7 +139,7 @@ async def generate_audio_via_podcast_api(text, speaker_name, output_path):
|
|
| 132 |
|
| 133 |
return await asyncio.to_thread(_sync_request)
|
| 134 |
|
| 135 |
-
# پردازش تکتک قطعات زیرنویس
|
| 136 |
async def process_single_subtitle(i, sub, temp_dir, sem):
|
| 137 |
async with sem:
|
| 138 |
assigned_speaker = sub.get('speaker_id', 'Charon')
|
|
@@ -164,7 +171,7 @@ async def process_single_subtitle(i, sub, temp_dir, sem):
|
|
| 164 |
# ==========================================
|
| 165 |
# هسته اصلی: کارگردان هوشمند
|
| 166 |
# ==========================================
|
| 167 |
-
async def process_dubbing(api_key, video_file, youtube_url, target_lang,
|
| 168 |
if not api_key: raise gr.Error("کلید API جمینای الزامی است.")
|
| 169 |
if not video_file and not youtube_url: raise gr.Error("ویدیو الزامی است.")
|
| 170 |
|
|
@@ -172,90 +179,122 @@ async def process_dubbing(api_key, video_file, youtube_url, target_lang, keep_or
|
|
| 172 |
|
| 173 |
try:
|
| 174 |
client = genai.Client(http_options={"api_version": "v1beta"}, api_key=api_key)
|
| 175 |
-
|
| 176 |
video_path = os.path.join(temp_dir, "input.mp4")
|
| 177 |
audio_path = os.path.join(temp_dir, "source.mp3")
|
| 178 |
final_path = os.path.join(temp_dir, "output.mp4")
|
| 179 |
-
|
| 180 |
progress(0.05, desc="دریافت ویدیو...")
|
| 181 |
-
if youtube_url:
|
| 182 |
-
download_youtube_video(youtube_url, video_path)
|
| 183 |
else:
|
| 184 |
s_path = video_file.name if hasattr(video_file, 'name') else str(video_file)
|
| 185 |
shutil.copy(s_path, video_path)
|
| 186 |
-
|
| 187 |
duration = get_video_duration(video_path)
|
| 188 |
-
|
| 189 |
progress(0.1, desc="استخراج صدا...")
|
| 190 |
extract_audio_from_video(video_path, audio_path)
|
| 191 |
-
|
| 192 |
-
|
|
|
|
| 193 |
gemini_file = client.files.upload(file=audio_path)
|
| 194 |
-
|
| 195 |
prompt = f"""
|
| 196 |
ROLE: You are an expert Dubbing Director using AI Voice Actors.
|
|
|
|
| 197 |
{CAST_PROMPT}
|
|
|
|
| 198 |
TASK:
|
| 199 |
-
1. Listen to the audio and identify different speakers
|
| 200 |
-
2.
|
| 201 |
3. Translate the text EXACTLY to {target_lang}.
|
| 202 |
4. Output JSON Array.
|
|
|
|
| 203 |
JSON FORMAT:
|
| 204 |
[
|
| 205 |
-
{{
|
| 206 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 207 |
]
|
| 208 |
"""
|
|
|
|
| 209 |
transcription = client.models.generate_content(
|
| 210 |
model='gemini-2.5-flash',
|
| 211 |
contents=[gemini_file, prompt],
|
| 212 |
config=types.GenerateContentConfig(response_mime_type="application/json")
|
| 213 |
)
|
|
|
|
| 214 |
try: client.files.delete(name=gemini_file.name)
|
| 215 |
except: pass
|
| 216 |
-
|
| 217 |
json_clean = transcription.text.strip().replace("```json", "").replace("```", "")
|
| 218 |
subtitles = json.loads(json_clean)
|
| 219 |
-
|
| 220 |
if not subtitles: raise ValueError("زیرنویس خالی است")
|
| 221 |
|
| 222 |
-
# ---
|
| 223 |
-
|
| 224 |
-
|
| 225 |
-
background_audio = AudioSegment.from_file(audio_path)
|
| 226 |
-
# تبدیل مقیاس 0-1 به دسیبل (dB) برای کاهش صدا
|
| 227 |
-
# کاهش 50 دسیبلی تقریباً معادل سکوت است.
|
| 228 |
-
volume_reduction_db = (1.0 - original_audio_volume) * 50
|
| 229 |
-
final_track = background_audio - volume_reduction_db
|
| 230 |
-
else:
|
| 231 |
-
# ایجاد یک فایل صوتی کاملاً ساکت با طول ویدیو
|
| 232 |
-
final_track = AudioSegment.silent(duration=int(duration * 1000))
|
| 233 |
-
|
| 234 |
total = len(subtitles)
|
| 235 |
ok_cnt = 0
|
| 236 |
|
| 237 |
sem = asyncio.Semaphore(20)
|
| 238 |
tasks = [process_single_subtitle(i, sub, temp_dir, sem) for i, sub in enumerate(subtitles)]
|
|
|
|
| 239 |
completed = 0
|
| 240 |
for coro in asyncio.as_completed(tasks):
|
| 241 |
res = await coro
|
| 242 |
completed += 1
|
| 243 |
-
progress(0.3 + (0.
|
| 244 |
-
|
| 245 |
if res is not None:
|
| 246 |
seg = AudioSegment.from_file(res["adj_p"])
|
| 247 |
-
|
|
|
|
| 248 |
ok_cnt += 1
|
| 249 |
|
| 250 |
if ok_cnt == 0: raise gr.Error("خطا: صدایی تولید نشد.")
|
| 251 |
|
| 252 |
-
|
| 253 |
-
|
| 254 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 255 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 256 |
cmd = ['ffmpeg', '-y', '-i', video_path, '-i', final_audio_p, '-c:v', 'copy', '-c:a', 'aac', '-map', '0:v:0', '-map', '1:a:0', final_path]
|
| 257 |
subprocess.run(cmd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, check=True)
|
| 258 |
-
|
| 259 |
return final_path, json.dumps(subtitles, ensure_ascii=False, indent=2)
|
| 260 |
|
| 261 |
except Exception as e:
|
|
@@ -267,31 +306,28 @@ async def process_dubbing(api_key, video_file, youtube_url, target_lang, keep_or
|
|
| 267 |
with gr.Blocks(title="AI Smart Director Dubbing", theme=gr.themes.Soft()) as app:
|
| 268 |
gr.Markdown("""
|
| 269 |
# 🎬 استودیو دوبله هوشمند (AI Director)
|
| 270 |
-
**قابلیت
|
| 271 |
""")
|
| 272 |
-
|
| 273 |
with gr.Row():
|
| 274 |
with gr.Column():
|
| 275 |
-
api_key = gr.Textbox(label="کلید API جمینای", type="password")
|
| 276 |
vid = gr.Video(label="فایل ویدیو")
|
| 277 |
url = gr.Textbox(label="لینک یوتیوب")
|
| 278 |
lang = gr.Dropdown(["Persian (فارسی)", "English", "Arabic"], value="Persian (فارسی)", label="زبان مقصد")
|
| 279 |
|
| 280 |
-
|
| 281 |
-
|
| 282 |
-
|
| 283 |
-
|
|
|
|
| 284 |
btn = gr.Button("🚀 شروع دوبله هوشمند", variant="primary")
|
| 285 |
-
|
| 286 |
with gr.Column():
|
| 287 |
out_vid = gr.Video(label="ویدیو خروجی")
|
| 288 |
-
out_log = gr.Code(label="گزارش کستینگ (
|
| 289 |
|
| 290 |
-
btn.click(
|
| 291 |
-
process_dubbing,
|
| 292 |
-
[api_key, vid, url, lang, keep_audio_checkbox, audio_volume_slider],
|
| 293 |
-
[out_vid, out_log]
|
| 294 |
-
)
|
| 295 |
|
| 296 |
if __name__ == "__main__":
|
| 297 |
app.launch(ssr_mode=False)
|
|
|
|
| 6 |
import shutil
|
| 7 |
import asyncio
|
| 8 |
import requests
|
| 9 |
+
import math
|
| 10 |
from pydub import AudioSegment
|
| 11 |
import yt_dlp
|
| 12 |
from google import genai
|
|
|
|
| 62 |
return output_path
|
| 63 |
|
| 64 |
def extract_audio_from_video(video_path, audio_path):
|
| 65 |
+
# استخراج صدا با فرمت mp3 برای پردازش
|
| 66 |
+
subprocess.run(['ffmpeg', '-i', video_path, '-vn', '-acodec', 'mp3', '-ar', '44100', '-ac', '2', '-b:a', '192k', '-y', audio_path], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, check=True)
|
| 67 |
return audio_path
|
| 68 |
|
| 69 |
def get_video_duration(video_path):
|
| 70 |
result = subprocess.run(['ffprobe', '-v', 'error', '-show_entries', 'format=duration', '-of', 'default=noprint_wrappers=1:nokey=1', video_path], capture_output=True, text=True, check=True)
|
| 71 |
return float(result.stdout.strip())
|
| 72 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 73 |
def adjust_audio_speed_ffmpeg(input_wav, output_wav, target_duration):
|
| 74 |
try:
|
| 75 |
audio = AudioSegment.from_file(input_wav)
|
| 76 |
+
# حذف سکوت اضافه
|
| 77 |
+
if len(audio) > 100:
|
| 78 |
+
def trim(snd):
|
| 79 |
+
start_trim = 0
|
| 80 |
+
end_trim = 0
|
| 81 |
+
silence_thresh = -45.0
|
| 82 |
+
chunk = 10
|
| 83 |
+
while start_trim < len(snd) and snd[start_trim:start_trim+chunk].dBFS < silence_thresh: start_trim += chunk
|
| 84 |
+
while end_trim < len(snd) and snd[len(snd)-end_trim-chunk:len(snd)-end_trim].dBFS < silence_thresh: end_trim += chunk
|
| 85 |
+
return snd[start_trim:len(snd)-end_trim]
|
| 86 |
+
audio = trim(audio)
|
| 87 |
+
|
| 88 |
+
if len(audio) < 50:
|
| 89 |
+
shutil.copy(input_wav, output_wav)
|
| 90 |
+
return
|
| 91 |
|
| 92 |
+
temp_trimmed = input_wav.replace(".wav", "_trimmed.wav")
|
| 93 |
+
audio.export(temp_trimmed, format="wav")
|
| 94 |
+
|
| 95 |
+
orig_dur = len(audio) / 1000.0
|
| 96 |
if orig_dur <= 0.1 or target_duration <= 0.1:
|
| 97 |
shutil.copy(temp_trimmed, output_wav); return
|
| 98 |
+
|
| 99 |
speed_factor = orig_dur / target_duration
|
| 100 |
+
# محدودیت سرعت برای جلوگیری از رباتی شدن بیش از حد
|
| 101 |
if speed_factor > 2.0: speed_factor = 2.0
|
| 102 |
if speed_factor < 0.6: speed_factor = 0.6
|
| 103 |
+
|
| 104 |
atempo_filters = []
|
| 105 |
current = speed_factor
|
| 106 |
while current > 2.0: atempo_filters.append("atempo=2.0"); current /= 2.0
|
| 107 |
while current < 0.5: atempo_filters.append("atempo=0.5"); current /= 0.5
|
| 108 |
if current != 1.0: atempo_filters.append(f"atempo={current}")
|
| 109 |
+
|
| 110 |
if not atempo_filters: shutil.copy(temp_trimmed, output_wav); return
|
| 111 |
+
|
| 112 |
subprocess.run(['ffmpeg', '-y', '-i', temp_trimmed, '-filter:a', ",".join(atempo_filters), output_wav], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, check=True)
|
| 113 |
except: shutil.copy(input_wav, output_wav)
|
| 114 |
|
|
|
|
| 139 |
|
| 140 |
return await asyncio.to_thread(_sync_request)
|
| 141 |
|
| 142 |
+
# پردازش تکتک قطعات زیرنویس
|
| 143 |
async def process_single_subtitle(i, sub, temp_dir, sem):
|
| 144 |
async with sem:
|
| 145 |
assigned_speaker = sub.get('speaker_id', 'Charon')
|
|
|
|
| 171 |
# ==========================================
|
| 172 |
# هسته اصلی: کارگردان هوشمند
|
| 173 |
# ==========================================
|
| 174 |
+
async def process_dubbing(api_key, video_file, youtube_url, target_lang, keep_bg_audio, bg_audio_volume, progress=gr.Progress()):
|
| 175 |
if not api_key: raise gr.Error("کلید API جمینای الزامی است.")
|
| 176 |
if not video_file and not youtube_url: raise gr.Error("ویدیو الزامی است.")
|
| 177 |
|
|
|
|
| 179 |
|
| 180 |
try:
|
| 181 |
client = genai.Client(http_options={"api_version": "v1beta"}, api_key=api_key)
|
| 182 |
+
|
| 183 |
video_path = os.path.join(temp_dir, "input.mp4")
|
| 184 |
audio_path = os.path.join(temp_dir, "source.mp3")
|
| 185 |
final_path = os.path.join(temp_dir, "output.mp4")
|
| 186 |
+
|
| 187 |
progress(0.05, desc="دریافت ویدیو...")
|
| 188 |
+
if youtube_url: download_youtube_video(youtube_url, video_path)
|
|
|
|
| 189 |
else:
|
| 190 |
s_path = video_file.name if hasattr(video_file, 'name') else str(video_file)
|
| 191 |
shutil.copy(s_path, video_path)
|
| 192 |
+
|
| 193 |
duration = get_video_duration(video_path)
|
| 194 |
+
|
| 195 |
progress(0.1, desc="استخراج صدا...")
|
| 196 |
extract_audio_from_video(video_path, audio_path)
|
| 197 |
+
|
| 198 |
+
# --- تحلیل هوش مصنوعی ---
|
| 199 |
+
progress(0.2, desc="تحلیل ویدیو و کستینگ...")
|
| 200 |
gemini_file = client.files.upload(file=audio_path)
|
| 201 |
+
|
| 202 |
prompt = f"""
|
| 203 |
ROLE: You are an expert Dubbing Director using AI Voice Actors.
|
| 204 |
+
|
| 205 |
{CAST_PROMPT}
|
| 206 |
+
|
| 207 |
TASK:
|
| 208 |
+
1. Listen to the audio and identify different speakers.
|
| 209 |
+
2. Assign the BEST MATCH from 'AVAILABLE VOICE ACTORS'.
|
| 210 |
3. Translate the text EXACTLY to {target_lang}.
|
| 211 |
4. Output JSON Array.
|
| 212 |
+
|
| 213 |
JSON FORMAT:
|
| 214 |
[
|
| 215 |
+
{{
|
| 216 |
+
"start": 0.0,
|
| 217 |
+
"end": 3.5,
|
| 218 |
+
"speaker_id": "Charon",
|
| 219 |
+
"original_speaker_desc": "Male, Deep voice",
|
| 220 |
+
"text": "ترجمه متن"
|
| 221 |
+
}}
|
| 222 |
]
|
| 223 |
"""
|
| 224 |
+
|
| 225 |
transcription = client.models.generate_content(
|
| 226 |
model='gemini-2.5-flash',
|
| 227 |
contents=[gemini_file, prompt],
|
| 228 |
config=types.GenerateContentConfig(response_mime_type="application/json")
|
| 229 |
)
|
| 230 |
+
|
| 231 |
try: client.files.delete(name=gemini_file.name)
|
| 232 |
except: pass
|
| 233 |
+
|
| 234 |
json_clean = transcription.text.strip().replace("```json", "").replace("```", "")
|
| 235 |
subtitles = json.loads(json_clean)
|
| 236 |
+
|
| 237 |
if not subtitles: raise ValueError("زیرنویس خالی است")
|
| 238 |
|
| 239 |
+
# --- تولید صدا ---
|
| 240 |
+
# ایجاد ترک خالی برای صداهای دوبله (Voices Only)
|
| 241 |
+
voice_track = AudioSegment.silent(duration=int(duration * 1000))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 242 |
total = len(subtitles)
|
| 243 |
ok_cnt = 0
|
| 244 |
|
| 245 |
sem = asyncio.Semaphore(20)
|
| 246 |
tasks = [process_single_subtitle(i, sub, temp_dir, sem) for i, sub in enumerate(subtitles)]
|
| 247 |
+
|
| 248 |
completed = 0
|
| 249 |
for coro in asyncio.as_completed(tasks):
|
| 250 |
res = await coro
|
| 251 |
completed += 1
|
| 252 |
+
progress(0.3 + (0.5 * (completed / total)), desc=f"تولید صداها ({completed} از {total})...")
|
| 253 |
+
|
| 254 |
if res is not None:
|
| 255 |
seg = AudioSegment.from_file(res["adj_p"])
|
| 256 |
+
# قرار دادن صدای دوبله روی ترک خالی
|
| 257 |
+
voice_track = voice_track.overlay(seg, position=int(res["start"] * 1000))
|
| 258 |
ok_cnt += 1
|
| 259 |
|
| 260 |
if ok_cnt == 0: raise gr.Error("خطا: صدایی تولید نشد.")
|
| 261 |
|
| 262 |
+
# --- میکس صدای پسزمینه (Duck / Mix) ---
|
| 263 |
+
progress(0.90, desc="میکس صدای پسزمینه...")
|
| 264 |
+
|
| 265 |
+
original_audio = AudioSegment.from_file(audio_path)
|
| 266 |
+
|
| 267 |
+
# اطمینان از هماندازه بودن ترکها (گاهی ffmpeg دقیق نیست)
|
| 268 |
+
if len(voice_track) > len(original_audio):
|
| 269 |
+
original_audio = original_audio + AudioSegment.silent(duration=len(voice_track)-len(original_audio))
|
| 270 |
+
else:
|
| 271 |
+
original_audio = original_audio[:len(voice_track)]
|
| 272 |
|
| 273 |
+
if keep_bg_audio:
|
| 274 |
+
if bg_audio_volume <= 0.001:
|
| 275 |
+
# اگر ولوم صفر بود، سکوت مطلق
|
| 276 |
+
bg_music = AudioSegment.silent(duration=len(original_audio))
|
| 277 |
+
else:
|
| 278 |
+
# تبدیل درصد ولوم (0 تا 1) به دسیبل
|
| 279 |
+
# فرمول: 20 * log10(ratio)
|
| 280 |
+
# مثال: 0.1 (10%) حدودا -20dB میشود
|
| 281 |
+
gain_db = 20 * math.log10(max(bg_audio_volume, 0.0001))
|
| 282 |
+
bg_music = original_audio + gain_db
|
| 283 |
+
|
| 284 |
+
# ترکیب صدای دوبله روی صدای پسزمینه
|
| 285 |
+
final_mix = bg_music.overlay(voice_track)
|
| 286 |
+
else:
|
| 287 |
+
# فقط صدای دوبله
|
| 288 |
+
final_mix = voice_track
|
| 289 |
+
|
| 290 |
+
final_audio_p = os.path.join(temp_dir, "final_mix.wav")
|
| 291 |
+
final_mix.export(final_audio_p, format="wav")
|
| 292 |
+
|
| 293 |
+
# --- چسباندن صدا به تصویر ---
|
| 294 |
+
progress(0.95, desc="رندر نهایی ویدیو...")
|
| 295 |
cmd = ['ffmpeg', '-y', '-i', video_path, '-i', final_audio_p, '-c:v', 'copy', '-c:a', 'aac', '-map', '0:v:0', '-map', '1:a:0', final_path]
|
| 296 |
subprocess.run(cmd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, check=True)
|
| 297 |
+
|
| 298 |
return final_path, json.dumps(subtitles, ensure_ascii=False, indent=2)
|
| 299 |
|
| 300 |
except Exception as e:
|
|
|
|
| 306 |
with gr.Blocks(title="AI Smart Director Dubbing", theme=gr.themes.Soft()) as app:
|
| 307 |
gr.Markdown("""
|
| 308 |
# 🎬 استودیو دوبله هوشمند (AI Director)
|
| 309 |
+
**قابلیتها:** کستینگ خودکار گویندهها، تنظیم صدای پسزمینه و سرعت بالا با پردازش موازی.
|
| 310 |
""")
|
| 311 |
+
|
| 312 |
with gr.Row():
|
| 313 |
with gr.Column():
|
| 314 |
+
api_key = gr.Textbox(label="کلید API جمینای (Gemini API Key)", type="password")
|
| 315 |
vid = gr.Video(label="فایل ویدیو")
|
| 316 |
url = gr.Textbox(label="لینک یوتیوب")
|
| 317 |
lang = gr.Dropdown(["Persian (فارسی)", "English", "Arabic"], value="Persian (فارسی)", label="زبان مقصد")
|
| 318 |
|
| 319 |
+
# --- تنظیمات جدید صدا ---
|
| 320 |
+
gr.Markdown("### 🎚️ تنظیمات صدا")
|
| 321 |
+
keep_bg = gr.Checkbox(label="حفظ صدای پسزمینه (Original Audio)", value=True)
|
| 322 |
+
bg_vol = gr.Slider(minimum=0.0, maximum=1.0, value=0.15, step=0.01, label="میزان صدای پسزمینه (0 = سکوت، 1 = صدای اصلی)")
|
| 323 |
+
|
| 324 |
btn = gr.Button("🚀 شروع دوبله هوشمند", variant="primary")
|
| 325 |
+
|
| 326 |
with gr.Column():
|
| 327 |
out_vid = gr.Video(label="ویدیو خروجی")
|
| 328 |
+
out_log = gr.Code(label="گزارش کستینگ (JSON)", language="json")
|
| 329 |
|
| 330 |
+
btn.click(process_dubbing, [api_key, vid, url, lang, keep_bg, bg_vol], [out_vid, out_log])
|
|
|
|
|
|
|
|
|
|
|
|
|
| 331 |
|
| 332 |
if __name__ == "__main__":
|
| 333 |
app.launch(ssr_mode=False)
|