Update main.py

main.py

@@ -1,12 +1,13 @@
 import tensorflow as tf
-import gradio as gr
+#from transformers import pipeline
 from huggingface_hub import from_pretrained_keras
 import pandas as pd
 import numpy as np
 import joblib
 import os
 import sys
-
+import pickle
+import shutil
 # librosa is a Python library for analyzing audio and music. It can be used to extract the data from the audio files we will see it later.
 import librosa
 import librosa.display
@@ -41,34 +42,42 @@ if not sys.warnoptions:
     warnings.simplefilter("ignore")
 warnings.filterwarnings("ignore", category=DeprecationWarning)
 
-model = from_pretrained_keras('Mohamed41/MODEL_EMOTION_AR_TEXT_72P')
-
-
-def feat_ext(data):
-    # Time_domain_features
-    # ZCR Persody features or Low level ascoustic features
-    result = np.array([])
-    zcr = np.mean(librosa.feature.zero_crossing_rate(y=data).T, axis=0)
-    result = np.hstack((result, zcr))  # stacking horizontally
-    # Frequency_domain_features
-    # Spectral and wavelet Features
-    # MFCC
-    mfcc = np.mean(librosa.feature.mfcc(y=data, sr=22050, n_mfcc=40).T, axis=0)
-    result = np.hstack((result, mfcc))  # stacking horizontally
+model=from_pretrained_keras( 'Mohamed41/MODEL_EMOTION_AR_TEXT_72P')
+
+  
+with open('scaler3.pickle', 'rb') as f:
+    scaler3 = pickle.load(f)
+    
+with open('encoder3.pickle', 'rb') as f:
+    encoder3 = pickle.load(f)
+
+def zcr(data,frame_length,hop_length):
+    zcr=librosa.feature.zero_crossing_rate(data,frame_length=frame_length,hop_length=hop_length)
+    return np.squeeze(zcr)
+def rmse(data,frame_length=2048,hop_length=512):
+    rmse=librosa.feature.rms(y=data,frame_length=frame_length,hop_length=hop_length)
+    return np.squeeze(rmse)
+def mfcc(data,sr,frame_length=2048,hop_length=512,flatten:bool=True):
+    mfcc=librosa.feature.mfcc(y=data,sr=sr)
+    return np.squeeze(mfcc.T)if not flatten else np.ravel(mfcc.T)
+
+def extract_features(data,sr=22050,frame_length=2048,hop_length=512):
+    result=np.array([])
+
+    result=np.hstack((result,
+                      zcr(data,frame_length,hop_length),
+                      rmse(data,frame_length,hop_length),
+                      mfcc(data,sr,frame_length,hop_length)
+                     ))
     return result
 
-
-scaler = joblib.load('scaler.joblib')
-encoder = joblib.load('encoder.joblib')
-
-
 def get_predict_feat(path):
-    d, s_rate = librosa.load(path, duration=2.5, offset=0.6)
-    res = feat_ext(d)
-    result = np.array(res)
-    result = np.reshape(result, newshape=(1, 41))
-    i_result = scaler.transform(result)
-    final_result = np.expand_dims(i_result, axis=2)
+    d, s_rate= librosa.load(path, duration=2.5, offset=0.6)
+    res=extract_features(d)
+    result=np.array(res)
+    result=np.reshape(result,newshape=(1,2376))
+    i_result = scaler3.transform(result)
+    final_result=np.expand_dims(i_result, axis=2)
 
     return final_result
 
@@ -78,17 +87,23 @@ emotions1 = {1: 'Neutral', 2: 'Calm', 3: 'Happy', 4: 'Sad',
 
 
 def prediction(path1):
-    res = get_predict_feat(path1)
-    predictions = model.predict(res)
-    y_pred = encoder.inverse_transform(predictions)
-    return y_pred[0][0]
+    res=get_predict_feat(path1)
+    predictions=loaded_model.predict(res)
+    y_pred = encoder3.inverse_transform(predictions)
+    print(y_pred[0][0])
 
 
 app = FastAPI()
 
 
-@app.post("/")
-async def read_root(request: Request, file: UploadFile = File(...)):
-    json_data = await request.json()
 
-    return {"filename": file.filename, "filepath": f"/app/{file.filename}"}
+
+@app.post("/")
+async def read_root( file: UploadFile = File(...)):
+    file_extension = os.path.splitext(file.filename)[1]
+    with open("tmp"+file_extension, "wb") as buffer:
+        shutil.copyfileobj(file.file, buffer)
+
+     
+    x = prediction("tmp"+file_extension) 
+    return {"filename": file.filename, "filepath": f"/app/{file.filename}","prediction":x}