Update app.py

app.py

@@ -1,29 +1,37 @@
 import streamlit as st
 from transformers import pipeline
+import torchaudio
+import os
+import ffmpeg
 
-# Load Whisper model (you can choose base, large, etc.)
-whisper = pipeline("automatic-speech-recognition", model="openai/whisper-base")
+# Step 1: Upload and convert audio to WAV format
+st.title("Audio Transcription with Whisper (Small Model)")
 
-# Streamlit app layout
-st.title("Audio Transcription with Whisper")
+uploaded_file = st.file_uploader("Upload an audio file", type=["mp3", "wav", "ogg", "opus"])
 
-st.write("Upload an audio file and let the model transcribe it into text.")
+if uploaded_file is not None:
+    # Save the uploaded file
+    audio_path = os.path.join("uploads", uploaded_file.name)
+    with open(audio_path, "wb") as f:
+        f.write(uploaded_file.getbuffer())
+    
+    st.audio(audio_path, format="audio/ogg")
 
-# Upload audio file
-audio_file = st.file_uploader("Upload an audio file", type=["mp3", "wav", "flac", "opus", "m4a"])
+    # Convert to WAV format using ffmpeg if it's not WAV already
+    wav_path = audio_path.replace(os.path.splitext(audio_path)[-1], ".wav")
+    if not audio_path.endswith(".wav"):
+        ffmpeg.input(audio_path).output(wav_path).run()
 
-if audio_file is not None:
-    # Save the uploaded audio file
-    with open("uploaded_audio.wav", "wb") as f:
-        f.write(audio_file.getbuffer())
+    st.success("Audio converted to WAV format")
 
-    # Display the audio file name
-    st.audio(audio_file, format="audio/wav")
-    
-    # Run transcription
+    # Step 2: Load Whisper small model for transcription
+    whisper = pipeline("automatic-speech-recognition", model="openai/whisper-small")
+
+    # Step 3: Transcribe the audio
     st.write("Transcribing...")
-    result = whisper("uploaded_audio.wav")["text"]
 
-    # Show the transcription result
-    st.subheader("Transcription:")
-    st.write(result)
+    transcription = whisper(wav_path)
+    st.write("Transcription result: ", transcription['text'])
+
+else:
+    st.info("Please upload an audio file to transcribe.")