Update app.py

app.py

@@ -1,14 +1,29 @@
 import streamlit as st
 import time
+import soundfile as sf
 from whisper_processor import process_audio
 
 def process_audio_streamlit(audio_file, model_name, lang):
-    start_time = time.time()
-    result = process_audio(audio_file, model_name=model_name, lang=lang)
-    end_time = time.time()
-    elapsed_time = end_time - start_time
-    st.write("Time taken:", elapsed_time, "seconds")
-    return result
+  start_time = time.time()
+
+  # Read audio data
+  audio, sample_rate = sf.read(audio_file)
+
+  # Check if conversion is necessary
+  if sample_rate != 16000:
+    # Resample to 16kHz
+    audio = sf.resample(audio, sample_rate, 16000)
+  
+  # Save the resampled audio (optional)
+  # sf.write("temp_resampled.wav", audio, 16000)
+
+  # Process the audio with Whisper
+  result = process_audio(audio, model_name=model_name, lang=lang)
+
+  end_time = time.time()
+  elapsed_time = end_time - start_time
+  st.write("Time taken:", elapsed_time, "seconds")
+  return result
 
 st.title("Audio Transcription")
 
@@ -20,13 +35,13 @@ model_name = st.selectbox("Select model", ["tiny", "base", "small", "medium", "l
 lang = st.selectbox("Select language", ["en", "hi", "fr", "de", "es", "it", "pt", "ru", "zh", "ja", "ko", "ar", "tr"])
 
 if uploaded_file is not None:
-    # Save the uploaded file to a temporary location
-    with open("temp.wav", "wb") as f:
-        f.write(uploaded_file.read())
+  # Save the uploaded file to a temporary location
+  with open("temp.wav", "wb") as f:
+    f.write(uploaded_file.read())
 
-    # Process the audio file
-    result = process_audio_streamlit("temp.wav", model_name, lang)
+  # Process the audio file
+  result = process_audio_streamlit("temp.wav", model_name, lang)
 
-    # Display the transcription result
-    st.write("Transcription:")
-    st.text_area("", value=result, height=300)
+  # Display the transcription result
+  st.write("Transcription:")
+  st.text_area("", value=result, height=300)