Update app.py

app.py

@@ -6,12 +6,25 @@ from huggingface_hub import login
 from scipy.io.wavfile import write
 import uuid
 import torch
+import boto3
 
 print("cuda", torch.cuda.is_available())
 
 access_token_read = os.environ.get('HF_TOKEN', None)
+access_key = os.environ.get('access_key', None)
+secret_access_key = os.environ.get('secret_access_key', None)
+
 login(token = access_token_read)
 
+session = boto3.Session(
+    aws_access_key_id=access_key,
+    aws_secret_access_key=secret_access_key,
+)
+
+s3 = session.resource('s3')
+
+BUCKET = "audio-text-938"
+
 print("cur path", os.listdir(os.path.join("..", "..", "..")))
 
 if not os.path.isdir(os.path.join("..", "..", "..", "data", "hfcache")):
@@ -40,10 +53,10 @@ def transcribe(audio):
     text = text.replace("à la ligne", "\n")
     text = text.replace("  virgule", ",")
     text = text.replace(" virgule", ",")
-    text = text.replace("  point", ".")
-    text = text.replace(" point", ".")
     text = text.replace("  deux points", ":")
     text = text.replace(" deux points", ":")
+    text = text.replace("  point", ".")
+    text = text.replace(" point", ".")
     text = text.replace(" nouveau paragraphe ", "\n\n")
     text = text.replace(" paragraphe ", "\n\n")
 
@@ -56,10 +69,13 @@ def save_fn(audio, text):
 
     uid = str(uuid.uuid4())
 
-    with open(os.path.join("..", "..", "..", "data", "audio_texts", f"{uid}.txt"), "w", encoding="utf-8") as f:
+    with open(f"{uid}.txt"), "w", encoding="utf-8") as f:
         f.write(text)
+    s3.Bucket(BUCKET).upload_file(f"{uid}.txt", f"texts/{uid}.txt") #local path, bucket path
 
-    write(os.path.join("..", "..", "..", "data", "audio", f"{uid}.wav"), sr, y)
+    write(f"{uid}.wav", sr, y)
+    s3.Bucket(BUCKET).upload_file(f"{uid}.wav", f"audios/{uid}.wav") #local path, bucket path
+    
     return [None, ""]
 
 with gr.Blocks() as demo: