Spaces:
Runtime error
Runtime error
Upload utils.py with huggingface_hub
Browse files
utils.py
ADDED
|
@@ -0,0 +1,232 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import tempfile
|
| 3 |
+
from pathlib import Path
|
| 4 |
+
import requests
|
| 5 |
+
|
| 6 |
+
def text_to_speech_gtts(text, language="fr"):
|
| 7 |
+
"""
|
| 8 |
+
Convertit du texte en audio en utilisant gTTS (Google Text-to-Speech)
|
| 9 |
+
|
| 10 |
+
Args:
|
| 11 |
+
text (str): Le texte à convertir
|
| 12 |
+
language (str): Code de langue (fr, en, es, etc.)
|
| 13 |
+
|
| 14 |
+
Returns:
|
| 15 |
+
str: Chemin vers le fichier audio généré
|
| 16 |
+
"""
|
| 17 |
+
try:
|
| 18 |
+
from gtts import gTTS
|
| 19 |
+
|
| 20 |
+
# Créer un fichier temporaire
|
| 21 |
+
output_path = tempfile.mktemp(suffix='.mp3')
|
| 22 |
+
|
| 23 |
+
# Générer l'audio
|
| 24 |
+
tts = gTTS(text=text, lang=language, slow=False)
|
| 25 |
+
tts.save(output_path)
|
| 26 |
+
|
| 27 |
+
return output_path
|
| 28 |
+
|
| 29 |
+
except ImportError:
|
| 30 |
+
raise ImportError("gTTS n'est pas installé. Installez-le avec: pip install gTTS")
|
| 31 |
+
except Exception as e:
|
| 32 |
+
raise Exception(f"Erreur lors de la génération TTS: {str(e)}")
|
| 33 |
+
|
| 34 |
+
def text_to_speech_elevenlabs(text, api_key, voice_id="EXAVITQu4vr4xnSDxMaL"):
|
| 35 |
+
"""
|
| 36 |
+
Convertit du texte en audio en utilisant ElevenLabs API
|
| 37 |
+
|
| 38 |
+
Args:
|
| 39 |
+
text (str): Le texte à convertir
|
| 40 |
+
api_key (str): Clé API ElevenLabs
|
| 41 |
+
voice_id (str): ID de la voix à utiliser
|
| 42 |
+
|
| 43 |
+
Returns:
|
| 44 |
+
str: Chemin vers le fichier audio généré
|
| 45 |
+
"""
|
| 46 |
+
try:
|
| 47 |
+
url = f"https://api.elevenlabs.io/v1/text-to-speech/{voice_id}"
|
| 48 |
+
|
| 49 |
+
headers = {
|
| 50 |
+
"Accept": "audio/mpeg",
|
| 51 |
+
"Content-Type": "application/json",
|
| 52 |
+
"xi-api-key": api_key
|
| 53 |
+
}
|
| 54 |
+
|
| 55 |
+
data = {
|
| 56 |
+
"text": text,
|
| 57 |
+
"model_id": "eleven_monolingual_v1",
|
| 58 |
+
"voice_settings": {
|
| 59 |
+
"stability": 0.5,
|
| 60 |
+
"similarity_boost": 0.5
|
| 61 |
+
}
|
| 62 |
+
}
|
| 63 |
+
|
| 64 |
+
response = requests.post(url, json=data, headers=headers)
|
| 65 |
+
|
| 66 |
+
if response.status_code == 200:
|
| 67 |
+
output_path = tempfile.mktemp(suffix='.mp3')
|
| 68 |
+
with open(output_path, 'wb') as f:
|
| 69 |
+
f.write(response.content)
|
| 70 |
+
return output_path
|
| 71 |
+
else:
|
| 72 |
+
raise Exception(f"Erreur API ElevenLabs: {response.status_code} - {response.text}")
|
| 73 |
+
|
| 74 |
+
except Exception as e:
|
| 75 |
+
raise Exception(f"Erreur lors de la génération TTS ElevenLabs: {str(e)}")
|
| 76 |
+
|
| 77 |
+
def merge_videos_side_by_side(video1_path, video2_path, output_path=None):
|
| 78 |
+
"""
|
| 79 |
+
Fusionne deux vidéos côte à côte
|
| 80 |
+
|
| 81 |
+
Args:
|
| 82 |
+
video1_path (str): Chemin vers la première vidéo
|
| 83 |
+
video2_path (str): Chemin vers la deuxième vidéo
|
| 84 |
+
output_path (str): Chemin de sortie (optionnel)
|
| 85 |
+
|
| 86 |
+
Returns:
|
| 87 |
+
str: Chemin vers la vidéo fusionnée
|
| 88 |
+
"""
|
| 89 |
+
import cv2
|
| 90 |
+
|
| 91 |
+
if output_path is None:
|
| 92 |
+
output_path = tempfile.mktemp(suffix='.mp4')
|
| 93 |
+
|
| 94 |
+
# Ouvrir les vidéos
|
| 95 |
+
cap1 = cv2.VideoCapture(video1_path)
|
| 96 |
+
cap2 = cv2.VideoCapture(video2_path)
|
| 97 |
+
|
| 98 |
+
# Obtenir les propriétés
|
| 99 |
+
fps = int(cap1.get(cv2.CAP_PROP_FPS))
|
| 100 |
+
height = int(cap1.get(cv2.CAP_PROP_FRAME_HEIGHT))
|
| 101 |
+
width1 = int(cap1.get(cv2.CAP_PROP_FRAME_WIDTH))
|
| 102 |
+
width2 = int(cap2.get(cv2.CAP_PROP_FRAME_WIDTH))
|
| 103 |
+
|
| 104 |
+
# Créer le writer pour la vidéo de sortie
|
| 105 |
+
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
|
| 106 |
+
out = cv2.VideoWriter(output_path, fourcc, fps, (width1 + width2, height))
|
| 107 |
+
|
| 108 |
+
while True:
|
| 109 |
+
ret1, frame1 = cap1.read()
|
| 110 |
+
ret2, frame2 = cap2.read()
|
| 111 |
+
|
| 112 |
+
if not ret1 or not ret2:
|
| 113 |
+
break
|
| 114 |
+
|
| 115 |
+
# Redimensionner si nécessaire pour avoir la même hauteur
|
| 116 |
+
if frame1.shape[0] != frame2.shape[0]:
|
| 117 |
+
frame2 = cv2.resize(frame2, (width2, height))
|
| 118 |
+
|
| 119 |
+
# Fusionner horizontalement
|
| 120 |
+
merged = cv2.hconcat([frame1, frame2])
|
| 121 |
+
out.write(merged)
|
| 122 |
+
|
| 123 |
+
cap1.release()
|
| 124 |
+
cap2.release()
|
| 125 |
+
out.release()
|
| 126 |
+
|
| 127 |
+
return output_path
|
| 128 |
+
|
| 129 |
+
def get_video_duration(video_path):
|
| 130 |
+
"""
|
| 131 |
+
Obtient la durée d'une vidéo en secondes
|
| 132 |
+
|
| 133 |
+
Args:
|
| 134 |
+
video_path (str): Chemin vers la vidéo
|
| 135 |
+
|
| 136 |
+
Returns:
|
| 137 |
+
float: Durée en secondes
|
| 138 |
+
"""
|
| 139 |
+
import cv2
|
| 140 |
+
|
| 141 |
+
cap = cv2.VideoCapture(video_path)
|
| 142 |
+
fps = cap.get(cv2.CAP_PROP_FPS)
|
| 143 |
+
frame_count = cap.get(cv2.CAP_PROP_FRAME_COUNT)
|
| 144 |
+
duration = frame_count / fps if fps > 0 else 0
|
| 145 |
+
cap.release()
|
| 146 |
+
|
| 147 |
+
return duration
|
| 148 |
+
|
| 149 |
+
def resize_video(input_path, output_path=None, width=None, height=None, scale=None):
|
| 150 |
+
"""
|
| 151 |
+
Redimensionne une vidéo
|
| 152 |
+
|
| 153 |
+
Args:
|
| 154 |
+
input_path (str): Chemin vers la vidéo source
|
| 155 |
+
output_path (str): Chemin de sortie (optionnel)
|
| 156 |
+
width (int): Nouvelle largeur (optionnel)
|
| 157 |
+
height (int): Nouvelle hauteur (optionnel)
|
| 158 |
+
scale (float): Facteur d'échelle (optionnel)
|
| 159 |
+
|
| 160 |
+
Returns:
|
| 161 |
+
str: Chemin vers la vidéo redimensionnée
|
| 162 |
+
"""
|
| 163 |
+
import cv2
|
| 164 |
+
|
| 165 |
+
if output_path is None:
|
| 166 |
+
output_path = tempfile.mktemp(suffix='.mp4')
|
| 167 |
+
|
| 168 |
+
cap = cv2.VideoCapture(input_path)
|
| 169 |
+
fps = int(cap.get(cv2.CAP_PROP_FPS))
|
| 170 |
+
orig_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
|
| 171 |
+
orig_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
|
| 172 |
+
|
| 173 |
+
# Calculer les nouvelles dimensions
|
| 174 |
+
if scale:
|
| 175 |
+
width = int(orig_width * scale)
|
| 176 |
+
height = int(orig_height * scale)
|
| 177 |
+
elif width and not height:
|
| 178 |
+
height = int(orig_height * (width / orig_width))
|
| 179 |
+
elif height and not width:
|
| 180 |
+
width = int(orig_width * (height / orig_height))
|
| 181 |
+
elif not width and not height:
|
| 182 |
+
width, height = orig_width, orig_height
|
| 183 |
+
|
| 184 |
+
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
|
| 185 |
+
out = cv2.VideoWriter(output_path, fourcc, fps, (width, height))
|
| 186 |
+
|
| 187 |
+
while True:
|
| 188 |
+
ret, frame = cap.read()
|
| 189 |
+
if not ret:
|
| 190 |
+
break
|
| 191 |
+
|
| 192 |
+
resized = cv2.resize(frame, (width, height))
|
| 193 |
+
out.write(resized)
|
| 194 |
+
|
| 195 |
+
cap.release()
|
| 196 |
+
out.release()
|
| 197 |
+
|
| 198 |
+
return output_path
|
| 199 |
+
|
| 200 |
+
def create_solid_color_video(width, height, duration, fps, color=(0, 0, 0), output_path=None):
|
| 201 |
+
"""
|
| 202 |
+
Crée une vidéo d'une couleur unie
|
| 203 |
+
|
| 204 |
+
Args:
|
| 205 |
+
width (int): Largeur
|
| 206 |
+
height (int): Hauteur
|
| 207 |
+
duration (float): Durée en secondes
|
| 208 |
+
fps (int): Images par seconde
|
| 209 |
+
color (tuple): Couleur BGR (default: noir)
|
| 210 |
+
output_path (str): Chemin de sortie (optionnel)
|
| 211 |
+
|
| 212 |
+
Returns:
|
| 213 |
+
str: Chemin vers la vidéo créée
|
| 214 |
+
"""
|
| 215 |
+
import cv2
|
| 216 |
+
import numpy as np
|
| 217 |
+
|
| 218 |
+
if output_path is None:
|
| 219 |
+
output_path = tempfile.mktemp(suffix='.mp4')
|
| 220 |
+
|
| 221 |
+
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
|
| 222 |
+
out = cv2.VideoWriter(output_path, fourcc, fps, (width, height))
|
| 223 |
+
|
| 224 |
+
frame = np.full((height, width, 3), color, dtype=np.uint8)
|
| 225 |
+
|
| 226 |
+
total_frames = int(duration * fps)
|
| 227 |
+
for _ in range(total_frames):
|
| 228 |
+
out.write(frame)
|
| 229 |
+
|
| 230 |
+
out.release()
|
| 231 |
+
|
| 232 |
+
return output_path
|