Update app.py

app.py

@@ -25,6 +25,7 @@ from replaceWords import replace_words
 from applyVad import apply_vad
 from wienerFilter import wiener_filter
 from highPassFilter import high_pass_filter
+from wavletDenoise import wavelet_denoise
 
 
 
@@ -74,45 +75,40 @@ def transcribe_hindi_old(audio):
     converted_text=text_to_int(replaced_words)
     return converted_text
 
-## implementation of noise reduction techniques.
 ###############################################
-def noise_reduction_pipeline(filepath):
-    # Your existing noise reduction code
-    audio, sr = librosa.load(filepath, sr=None)
-    audio_hp = high_pass_filter(audio, sr, cutoff=100, order=5)
-    audio_wiener = wiener_filter(audio_hp)
-    audio_vad = apply_vad(audio_wiener, sr)
-    output_filepath = "processed_output.wav"
-    sf.write(output_filepath, audio_vad, sr)
-    return output_filepath
-
-# Hugging Face ASR function uses the pre-loaded model
-def transcribe_with_huggingface(filepath):
-    result = transcriber_hindi_lm(filepath)
-    text_value = result['text']
-    cleaned_text = text_value.replace("<s>", "")
-    converted_to_list = convert_to_list(cleaned_text, text_to_list())
-    processed_doubles = process_doubles(converted_to_list)
-    replaced_words = replace_words(processed_doubles)
-    converted_text = text_to_int(replaced_words)
-    print("Transcription: ", converted_text)
+# implementation of noise reduction techniques.
+
+# Function to apply a Wiener filter for noise reduction
+def apply_wiener_filter(audio):
+    return wiener(audio)
+
+# Function to handle speech recognition
+def Noise_cancellation_function(audio_file):
+    # Load the audio file using librosa
+    audio, sr = librosa.load(audio_file, sr=16000)
+
+    # Step 1: Apply a high-pass filter
+    audio = high_pass_filter(audio, sr)
+
+    # Step 2: Apply Wiener filter for noise reduction
+    audio = apply_wiener_filter(audio)
+
+    # Step 3: Apply wavelet denoising
+    denoised_audio = wavelet_denoise(audio)
+
+    # Save the denoised audio to a temporary file
+    temp_wav = "temp_denoised.wav"
+    write(temp_wav, sr, denoised_audio)
+
+    # Perform speech recognition on the denoised audio
+    transcript = transcriber_hindi_lm(temp_wav)
+    text_value = transcript['text']
+    cleaned_text=text_value.replace("<s>","")
+    processd_doubles=process_doubles(cleaned_text)
+    replaced_words = replace_words(processd_doubles)
+    converted_text=text_to_int(replaced_words)
     return converted_text
 
-# Combined function to process and transcribe audio
-def process_audio_and_transcribe(audio):
-    # Step 1: Preprocess (Noise Reduction)
-    try:
-        processed_filepath = noise_reduction_pipeline(audio)
-    except webrtcvad.Error as e:
-        return f"Error in processing audio for VAD: {str(e)}"
-    
-    # Step 2: Transcription
-    try:
-        transcription = transcribe_with_huggingface(processed_filepath)
-    except Exception as e:
-        return f"Transcription failed: {str(e)}"
-    
-    return transcription
 #################################################
 
 def sel_lng(lng, mic=None, file=None):
@@ -130,7 +126,7 @@ def sel_lng(lng, mic=None, file=None):
     elif lng== "model_3":
         return transcribe_hindi_lm(audio)
     elif lng== "model_4":
-        return process_audio_and_transcribe(audio)
+        return Noise_cancellation_function(audio)
             
         
 # demo=gr.Interface(