Spaces:
Paused
Paused
rick
commited on
bug fixed now... not done...
Browse files- pages/main.py +45 -15
pages/main.py
CHANGED
|
@@ -15,7 +15,6 @@ from typing import Optional
|
|
| 15 |
from typing import Tuple
|
| 16 |
from typing import Union
|
| 17 |
|
| 18 |
-
|
| 19 |
# Third-party libraries
|
| 20 |
import requests
|
| 21 |
import streamlit as st
|
|
@@ -91,9 +90,12 @@ def read_file(file_name: str) -> str:
|
|
| 91 |
except IOError as e:
|
| 92 |
return f"{get_translation('erreur_lecture_fichier')} {str(e)}"
|
| 93 |
|
| 94 |
-
|
|
|
|
|
|
|
|
|
|
| 95 |
"""
|
| 96 |
-
Divise un fichier audio en segments de
|
| 97 |
|
| 98 |
Args:
|
| 99 |
audio_file (str): Chemin vers le fichier audio.
|
|
@@ -111,13 +113,36 @@ def split_audio(audio_file: str, max_size_mb: int = 25) -> List[str]:
|
|
| 111 |
)
|
| 112 |
|
| 113 |
segments = []
|
| 114 |
-
|
| 115 |
-
|
| 116 |
-
|
| 117 |
-
|
| 118 |
-
|
| 119 |
-
|
| 120 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 121 |
return segments
|
| 122 |
except IOError as e:
|
| 123 |
print(f"Erreur lors de la lecture ou de l'écriture du fichier audio : {e}")
|
|
@@ -125,6 +150,9 @@ def split_audio(audio_file: str, max_size_mb: int = 25) -> List[str]:
|
|
| 125 |
except ValueError as e:
|
| 126 |
print(f"Erreur de valeur lors du traitement de l'audio : {e}")
|
| 127 |
return []
|
|
|
|
|
|
|
|
|
|
| 128 |
|
| 129 |
# Fonction modifiée pour transcrire l'audio en texte
|
| 130 |
def transcribe_audio(audio_file: IO, language: Optional[str] = None) -> str:
|
|
@@ -153,10 +181,10 @@ def transcribe_audio(audio_file: IO, language: Optional[str] = None) -> str:
|
|
| 153 |
transcript = client.audio.transcriptions.create(
|
| 154 |
model="whisper-1",
|
| 155 |
file=audio_segment,
|
| 156 |
-
language=language
|
|
|
|
| 157 |
)
|
| 158 |
-
full_transcript += f"{transcript
|
| 159 |
-
os.unlink(segment) # Supprime le fichier temporaire
|
| 160 |
status.update(label="Transcription terminée", state="complete")
|
| 161 |
return full_transcript.strip()
|
| 162 |
else:
|
|
@@ -165,16 +193,18 @@ def transcribe_audio(audio_file: IO, language: Optional[str] = None) -> str:
|
|
| 165 |
transcript = client.audio.transcriptions.create(
|
| 166 |
model="whisper-1",
|
| 167 |
file=audio_file,
|
| 168 |
-
language=language
|
|
|
|
| 169 |
)
|
| 170 |
status.update(label="Transcription terminée", state="complete")
|
| 171 |
-
return transcript
|
| 172 |
except IOError as e:
|
| 173 |
st.error(f"Erreur d'entrée/sortie lors de la transcription : {e}")
|
| 174 |
return ""
|
| 175 |
except Exception as e:
|
| 176 |
st.error(f"Erreur lors de la transcription : {e}")
|
| 177 |
return ""
|
|
|
|
| 178 |
|
| 179 |
# Fonction pour détecter la langue d'un texte donné
|
| 180 |
def detect_language(input_text: str, temperature: float = 0.01) -> str:
|
|
|
|
| 15 |
from typing import Tuple
|
| 16 |
from typing import Union
|
| 17 |
|
|
|
|
| 18 |
# Third-party libraries
|
| 19 |
import requests
|
| 20 |
import streamlit as st
|
|
|
|
| 90 |
except IOError as e:
|
| 91 |
return f"{get_translation('erreur_lecture_fichier')} {str(e)}"
|
| 92 |
|
| 93 |
+
|
| 94 |
+
|
| 95 |
+
def split_audio(audio_file: str,
|
| 96 |
+
max_size_mb: int = 25) -> List[str]:
|
| 97 |
"""
|
| 98 |
+
Divise un fichier audio en segments de taille maximale spécifiée et retourne les chemins des segments.
|
| 99 |
|
| 100 |
Args:
|
| 101 |
audio_file (str): Chemin vers le fichier audio.
|
|
|
|
| 113 |
)
|
| 114 |
|
| 115 |
segments = []
|
| 116 |
+
with tempfile.TemporaryDirectory() as temp_dir:
|
| 117 |
+
for start in range(0, duration_ms, segment_duration_ms):
|
| 118 |
+
end = min(start + segment_duration_ms, duration_ms)
|
| 119 |
+
segment = audio[start:end]
|
| 120 |
+
temp_segment_path = os.path.join(temp_dir, f"segment_{start}_{end}.wav")
|
| 121 |
+
segment.export(temp_segment_path, format="wav")
|
| 122 |
+
segments.append(temp_segment_path)
|
| 123 |
+
|
| 124 |
+
|
| 125 |
+
# Start of Selection
|
| 126 |
+
# Traiter les segments ici
|
| 127 |
+
language = st.session_state.get('target_language', 'en') # Définir la variable
|
| 128 |
+
transcriptions = []
|
| 129 |
+
for segment_path in segments:
|
| 130 |
+
try:
|
| 131 |
+
with open(segment_path, "rb") as audio_segment:
|
| 132 |
+
transcript = client.audio.transcriptions.create(
|
| 133 |
+
model="whisper-1",
|
| 134 |
+
file=audio_segment,
|
| 135 |
+
# Start of Selection
|
| 136 |
+
language=language,
|
| 137 |
+
response_format="text"
|
| 138 |
+
)
|
| 139 |
+
transcriptions.append(transcript.text)
|
| 140 |
+
except Exception as e:
|
| 141 |
+
print(f"Erreur lors de la transcription du segment {segment_path} : {e}")
|
| 142 |
+
finally:
|
| 143 |
+
os.remove(segment_path) # Supprime le segment après traitement
|
| 144 |
+
# Fin du traitement des segments et suppression automatique du répertoire temporaire
|
| 145 |
+
|
| 146 |
return segments
|
| 147 |
except IOError as e:
|
| 148 |
print(f"Erreur lors de la lecture ou de l'écriture du fichier audio : {e}")
|
|
|
|
| 150 |
except ValueError as e:
|
| 151 |
print(f"Erreur de valeur lors du traitement de l'audio : {e}")
|
| 152 |
return []
|
| 153 |
+
except Exception as e:
|
| 154 |
+
print(f"Une erreur inattendue s'est produite : {e}")
|
| 155 |
+
return []
|
| 156 |
|
| 157 |
# Fonction modifiée pour transcrire l'audio en texte
|
| 158 |
def transcribe_audio(audio_file: IO, language: Optional[str] = None) -> str:
|
|
|
|
| 181 |
transcript = client.audio.transcriptions.create(
|
| 182 |
model="whisper-1",
|
| 183 |
file=audio_segment,
|
| 184 |
+
language=language,
|
| 185 |
+
response_format="text"
|
| 186 |
)
|
| 187 |
+
full_transcript += f"{transcript} "
|
|
|
|
| 188 |
status.update(label="Transcription terminée", state="complete")
|
| 189 |
return full_transcript.strip()
|
| 190 |
else:
|
|
|
|
| 193 |
transcript = client.audio.transcriptions.create(
|
| 194 |
model="whisper-1",
|
| 195 |
file=audio_file,
|
| 196 |
+
language=language,
|
| 197 |
+
response_format="text"
|
| 198 |
)
|
| 199 |
status.update(label="Transcription terminée", state="complete")
|
| 200 |
+
return transcript
|
| 201 |
except IOError as e:
|
| 202 |
st.error(f"Erreur d'entrée/sortie lors de la transcription : {e}")
|
| 203 |
return ""
|
| 204 |
except Exception as e:
|
| 205 |
st.error(f"Erreur lors de la transcription : {e}")
|
| 206 |
return ""
|
| 207 |
+
## def transcribe_audio(audio_file: IO, language: Optional[str] = None) -> str:
|
| 208 |
|
| 209 |
# Fonction pour détecter la langue d'un texte donné
|
| 210 |
def detect_language(input_text: str, temperature: float = 0.01) -> str:
|