Update main.py

main.py

@@ -9,6 +9,7 @@ from typing import Union, Optional
 
 from fastapi import FastAPI, File, UploadFile, Form
 from fastapi.responses import StreamingResponse
+from fastapi.middleware.gzip import GZipMiddleware
 
 from utils import tts, read_image_file, pil_to_base64, base64_to_pil, get_hist
 from typing import Optional
@@ -25,7 +26,9 @@ CLASS = model.model.names
 defaul_bot_voice = "おはいようございます"
 area_thres = 0.3
 ZIP = False
+
 app = FastAPI()
+app.add_middleware(GZipMiddleware, minimum_size=1000)
 
 @app.get("/")
 def read_root():
@@ -61,6 +64,7 @@ async def predict_api(
                 out_img = image.crop(tuple(box)).resize((64, 64))
                 most_close = area_rate
     print("Get face time", time.time() - start_time)
+    
     start_time = time.time()
     if last_seen is not None:
         if type(last_seen) == str:
@@ -70,13 +74,14 @@ async def predict_api(
         if out_img is not None:
             diff_value = dist.euclidean(get_hist(out_img), get_hist(last_seen))
     print("Hist time", time.time() - start_time)
+    
     start_time = time.time()
     print(f"Distance: {most_close}. Different value: {diff_value}")
     if most_close >= area_thres and diff_value >= 0.5:
-        voice_bot_path = tts(defaul_bot_voice, language="ja")
-        image_bot_path = pil_to_base64(out_img)
-        print("Voice time", time.time() - start_time)
         if ZIP:
+            voice_bot_path = tts(defaul_bot_voice, language="ja")
+            image_bot_path = pil_to_base64(out_img)
+            print("Voice time", time.time() - start_time)
             io = BytesIO()
             zip_filename = "final_archive.zip"
             with zipfile.ZipFile(io, mode='w', compression=zipfile.ZIP_DEFLATED) as zf:
@@ -90,6 +95,9 @@ async def predict_api(
                 headers={"Content-Disposition": f"attachment;filename=%s" % zip_filename}
             )
         else:
+            voice_bot_path = tts(default_bot_voice, language="ja", encode=True)
+            image_bot_path = pil_to_base64(out_img, encode=True)
+            print("Voice time", time.time() - start_time)
             print("Total time", time.time() - total_time)
             return {
                 "voice": voice_bot_path,