Update app.py

app.py

@@ -3,30 +3,14 @@ import whisper
 import os
 from pydub import AudioSegment
 
-# Load a larger Whisper model for better accuracy
-model = whisper.load_model("medium")  # Use "medium" or "large" for better results
-
-def transcribe_audio(audio_file, language="Auto Detect"):
-    # Convert audio to 16kHz mono for better compatibility with Whisper
-    audio = AudioSegment.from_file(audio_file)
-    audio = audio.set_frame_rate(16000).set_channels(1)
-    processed_audio_path = "processed_audio.wav"
-    audio.export(processed_audio_path, format="wav")
-    
-    # Transcribe the audio
-    if language == "Auto Detect":
-        result = model.transcribe(processed_audio_path, fp16=False)  # Auto-detect language
-        detected_language = result.get("language", "unknown")
-    else:
-        language_code = LANGUAGE_NAME_TO_CODE.get(language, "en")  # Default to English if not found
-        result = model.transcribe(processed_audio_path, language=language_code, fp16=False)
-        detected_language = language_code
-    
-    # Clean up processed audio file
-    os.remove(processed_audio_path)
-    
-    # Return transcription and detected language
-    return f"Detected Language: {detected_language}\n\nTranscription:\n{result['text']}"
+# Mapping of model names to Whisper model sizes
+MODELS = {
+    "Tiny (Fastest)": "tiny",
+    "Base (Faster)": "base",
+    "Small (Balanced)": "small",
+    "Medium (Accurate)": "medium",
+    "Large (Most Accurate)": "large"
+}
 
 # Mapping of full language names to language codes
 LANGUAGE_NAME_TO_CODE = {
@@ -94,7 +78,7 @@ LANGUAGE_NAME_TO_CODE = {
     "Galician": "gl",
     "Marathi": "mr",
     "Punjabi": "pa",
-    "Sinhala": "si", 
+    "Sinhala": "si",  # Sinhala support
     "Khmer": "km",
     "Shona": "sn",
     "Yoruba": "yo",
@@ -132,6 +116,31 @@ LANGUAGE_NAME_TO_CODE = {
     "Sundanese": "su",
 }
 
+def transcribe_audio(audio_file, language="Auto Detect", model_size="Base (Faster)"):
+    # Load the selected Whisper model
+    model = whisper.load_model(MODELS[model_size])
+    
+    # Convert audio to 16kHz mono for better compatibility with Whisper
+    audio = AudioSegment.from_file(audio_file)
+    audio = audio.set_frame_rate(16000).set_channels(1)
+    processed_audio_path = "processed_audio.wav"
+    audio.export(processed_audio_path, format="wav")
+    
+    # Transcribe the audio
+    if language == "Auto Detect":
+        result = model.transcribe(processed_audio_path, fp16=False)  # Auto-detect language
+        detected_language = result.get("language", "unknown")
+    else:
+        language_code = LANGUAGE_NAME_TO_CODE.get(language, "en")  # Default to English if not found
+        result = model.transcribe(processed_audio_path, language=language_code, fp16=False)
+        detected_language = language_code
+    
+    # Clean up processed audio file
+    os.remove(processed_audio_path)
+    
+    # Return transcription and detected language
+    return f"Detected Language: {detected_language}\n\nTranscription:\n{result['text']}"
+
 # Define the Gradio interface
 iface = gr.Interface(
     fn=transcribe_audio,
@@ -141,11 +150,16 @@ iface = gr.Interface(
             choices=list(LANGUAGE_NAME_TO_CODE.keys()),  # Full language names
             label="Select Language",
             value="Auto Detect"
+        ),
+        gr.Dropdown(
+            choices=list(MODELS.keys()),  # Model options
+            label="Select Model",
+            value="Base (Faster)"  # Default to "Base" model
         )
     ],
     outputs=gr.Textbox(label="Transcription and Detected Language"),
-    title="Audio Transcription with Language Selection",
-    description="Upload an audio file and select a language (or choose 'Auto Detect')."
+    title="Audio Transcription with Language and Model Selection",
+    description="Upload an audio file, select a language (or choose 'Auto Detect'), and choose a model for transcription."
 )
 
 # Launch the Gradio interface