Create app.py

app.py

@@ -0,0 +1,43 @@
+import streamlit as st
+from faster_whisper import WhisperModel
+import logging
+
+# Configure logging for debugging purposes
+logging.basicConfig()
+logging.getLogger("faster_whisper").setLevel(logging.DEBUG)
+
+def format_timestamp(seconds):
+    """Convert seconds to HH:MM:SS.mmm format."""
+    hours = int(seconds // 3600)
+    minutes = int((seconds % 3600) // 60)
+    seconds_remainder = seconds % 60
+    return f"{hours:02d}:{minutes:02d}:{seconds_remainder:06.3f}"
+
+def transcribe(audio_file, model_size):
+    # Initialize the Whisper model based on the selected model size
+    device = "cpu"  # GPU : cuda  CPU : cpu
+    compute_type = "int8"  # GPU : float16 or int8 - CPU : int8
+    
+    model = WhisperModel(model_size, device=device, compute_type=compute_type)
+    
+    # Transcribe the audio file
+    segments, _ = model.transcribe(audio_file.name, audio_file)
+    
+    # Format and gather transcription with enhanced timestamps
+    transcription_with_timestamps = [
+        f"[{format_timestamp(segment.start)} -> {format_timestamp(segment.end)}] {segment.text}"
+        for segment in segments
+    ]
+    
+    return "\n".join(transcription_with_timestamps)
+
+# Streamlit UI
+st.title("Whisper API")
+st.write("For API use please visit [this space](https://huggingface.co/spaces/Lenylvt/Whisper-API)")
+
+audio_file = st.file_uploader("🎵 Upload Audio or Video", type=['wav', 'mp3', 'ogg', 'mp4', 'avi'])
+model_size = st.selectbox("📁 Model Size", ["base", "small", "medium", "large", "large-v2", "large-v3"])
+
+if audio_file is not None and model_size is not None:
+    transcription = transcribe(audio_file, model_size)
+    st.text_area("📜 Transcription", transcription, height=300)