modified for mp3 files too

app.py

@@ -5,6 +5,9 @@ import torchaudio
 import torch
 import os
 import soundfile as sf
+# New imports for handling MP3 and M4A files
+from pydub import AudioSegment
+import io
 
 model_name = "mispeech/ced-tiny"
 feature_extractor = CedFeatureExtractor.from_pretrained(model_name)
@@ -20,9 +23,17 @@ if audio_file is not None:
     st.write(f"Uploaded file: {audio_file.name}")
     
     try:
+        # New code block for handling different audio formats
         temp_file_path = "temp.wav"
-        with open(temp_file_path, "wb") as f:
-            f.write(audio_file.read())
+        if audio_file.name.lower().endswith(('.mp3', '.m4a')):
+            # Convert MP3/M4A to WAV
+            audio_bytes = audio_file.read()
+            audio = AudioSegment.from_file(io.BytesIO(audio_bytes), format=audio_file.name.split('.')[-1])
+            audio.export(temp_file_path, format="wav")
+        else:
+            # For WAV files, write directly
+            with open(temp_file_path, "wb") as f:
+                f.write(audio_file.read())
         
         try:
             audio, sampling_rate = torchaudio.load(temp_file_path)
@@ -51,4 +62,4 @@ if audio_file is not None:
     except Exception as e:
         st.error(f"An error occurred: {e}")
 else:
-    st.info("Please upload a .wav audio file to continue.")
+    st.info("Please upload an audio file (WAV, MP3, or M4A) to continue.")