Update app.py

app.py

@@ -24,11 +24,18 @@ else:
 
 model_id = "sudoping01/maliba-asr-v1"
 transcriber = WhosperTranscriber(model_id=model_id)
-logger.info(f"MALIBA-ASR-v1 transcriber initialized successfully")
+logger.info(f"Transcriber initialized with model: {model_id}")
 
 def resample_audio(audio_path, target_sample_rate=16000):
+
     """
     Converts the audio file to the target sampling rate (16000 Hz).
+    
+    Args:
+        audio_path (str): Path to the audio file.
+        target_sample_rate (int): The desired sample rate.
+    Returns:
+        A tensor containing the resampled audio data and the target sample rate.
     """
     try:
         waveform, original_sample_rate = torchaudio.load(audio_path)
@@ -47,15 +54,25 @@ def resample_audio(audio_path, target_sample_rate=16000):
 
 @spaces.GPU()
 def transcribe_audio(audio_file):
+
     """
     Transcribes the provided audio file into Bambara text using Whosper.
+    
+    Args:
+        audio_file: The path to the audio file to transcribe.
+    Returns:
+        A string representing the transcribed Bambara text.
     """
+
     if audio_file is None:
         return "Please provide an audio file for transcription."
     
     try:
         logger.info(f"Transcribing audio file: {audio_file}")
+        
+
         result = transcriber.transcribe_audio(audio_file)
+        
         logger.info("Transcription successful.")
         return result
         
@@ -64,14 +81,16 @@ def transcribe_audio(audio_file):
         return f"Error during transcription: {str(e)}"
 
 def get_example_files(directory="./examples"):
+
     """
-    Returns a list of audio files from the examples directory formatted for gr.Interface examples.
+    Returns a list of audio files from the examples directory.
     
     Args:
         directory (str): The directory to search for audio files.
     Returns:
-        list: A list of [audio_path] for each example file.
+        list: A list of paths to the audio files.
     """
+
     if not os.path.exists(directory):
         logger.warning(f"Examples directory {directory} not found.")
         return []
@@ -82,14 +101,10 @@ def get_example_files(directory="./examples"):
     
     try:
         files = os.listdir(directory)
-=
-        files.sort()
-        
         for file in files:
             if any(file.lower().endswith(ext) for ext in audio_extensions):
                 full_path = os.path.abspath(os.path.join(directory, file))
-   =
-                audio_files.append([full_path])
+                audio_files.append(full_path)
         
         logger.info(f"Found {len(audio_files)} example audio files.")
         return audio_files[:5]  
@@ -98,73 +113,127 @@ def get_example_files(directory="./examples"):
         logger.error(f"Error reading examples directory: {e}")
         return []
 
-def main():
+def build_interface():
     """
-    Main function to launch the Gradio interface using gr.Interface.
+    Builds the Gradio interface for Bambara speech recognition.
     """
-    logger.info("Starting MALIBA-ASR-v1 Gradio interface.")
-    
 
     example_files = get_example_files()
     
-
-    interface = gr.Interface(
-        fn=transcribe_audio,
-        inputs=[
-            gr.Audio(
-                label=" Record or Upload Audio", 
-                type="filepath",
-                sources=["microphone", "upload"]
+    with gr.Blocks(title="Bambara Speech Recognition") as demo:
+        gr.Markdown(
+            """
+            # 🎤 Bambara Automatic Speech Recognition
+            
+            **Powered by MALIBA-AI**
+            
+            Convert Bambara speech to text using our state-of-the-art ASR model. You can either:
+            - 🎙️ **Record** your voice directly
+            - 📁 **Upload** an audio file
+            - 🎵 **Try** our example audio files
+            
+            ## Supported Audio Formats
+            WAV, MP3, M4A, FLAC, OGG
+            """
+        )
+        
+        with gr.Row():
+            with gr.Column():
+
+                audio_input = gr.Audio(
+                    label="🎤 Record or Upload Audio",
+                    type="filepath",
+                    sources=["microphone", "upload"]
+                )
+                
+                transcribe_btn = gr.Button(
+                    "🔄 Transcribe Audio", 
+                    variant="primary",
+                    size="lg"
+                )
+                
+
+                clear_btn = gr.Button("🗑️ Clear", variant="secondary")
+            
+            with gr.Column():
+                output_text = gr.Textbox(
+                    label="📝 Transcribed Text (Bambara)",
+                    lines=8,
+                    placeholder="Your transcribed Bambara text will appear here...",
+                    interactive=False
+                )
+        
+        # Examples section
+        if example_files:
+            gr.Markdown("## 🎵 Try These Examples")
+            gr.Examples(
+                examples=[[f] for f in example_files],
+                inputs=[audio_input],
+                outputs=output_text,
+                fn=transcribe_audio,
+                cache_examples=False,
+                label="Example Audio Files"
             )
-        ],
-        outputs=gr.Textbox(
-            label="📝 Transcribed Text (Bambara)",
-            lines=8,
-            placeholder="Your transcribed Bambara text will appear here..."
-        ),
-        title="🎤 MALIBA-ASR-v1: Bambara Speech Recognition",
-        description="""
-                    **Revolutionizing Bambara Speech Technology | Powered by MALIBA-AI**
-                    
-                    Experience breakthrough Bambara speech recognition with **MALIBA-ASR-v1** - the most advanced open-source ASR model for Bambara, serving over 22 million speakers across Mali and West Africa.
-                    
-                    
-                    **Performance**: WER 0.226 | CER 0.109 on (6-hour test set)
-                            """,
-        examples=example_files if example_files else None,
-        cache_examples=False,
-        article="""
----
-
-## 🏆 MALIBA-ASR-v1 Performance
-
-| Metric | Value | Benchmark |
-|--------|-------|-----------|
-| **WER** | **0.226** | oza75/bambara-asr (test set) |
-| **CER** | **0.109** | oza75/bambara-asr (test set) |
-| **Test Duration** | **6 hours** | Diverse speakers & dialects |
-
-##  Revolutionary Impact
-
-**MALIBA-ASR-v1** sets a new standard for Bambara speech recognition, significantly outperforming all existing open-source solutions. This breakthrough enables:
-
-##  🇲🇱 About MALIBA-AI  🇲🇱
-
-MALIBA-AI is committed to ensuring **"No Malian Language Left Behind"** by:
-- Breaking digital language barriers for 22+ million Bambara speakers
-- Building cutting-edge AI technology for African languages
-- Preserving Mali's rich linguistic and cultural heritage
-- Democratizing access to voice technology across literacy levels
-- Training the next generation of African AI researchers
-
----
-
-**Model**: [sudoping01/maliba-asr-v1](https://huggingface.co/sudoping01/maliba-asr-v1) | **Dataset**: [oza75/bambara-asr](https://huggingface.co/datasets/oza75/bambara-asr)
-
-*Empowering Mali's Future Through Community-Driven AI Innovation* 🇲🇱
-        """
-    )
+        
+        # Information section
+        gr.Markdown(
+            """
+            ---
+            
+            ## ℹ️ About This Model
+            
+            - **Model:** [sudoping01/maliba-asr-v1](https://huggingface.co/sudoping01/maliba-asr-v1)
+            - **Developer:** MALIBA-AI
+            - **Language:** Bambara (bm)
+            - **Task:** Automatic Speech Recognition (ASR)
+            - **Sample Rate:** 16kHz (automatically resampled)
+            
+            ## 🚀 How to Use
+            
+            1. **Record Audio:** Click the microphone button and speak in Bambara
+            2. **Upload File:** Click the upload button to select an audio file
+            3. **Transcribe:** Click the "Transcribe Audio" button
+            4. **View Results:** See your transcribed text in Bambara
+            
+            ## 📊 Performance Notes
+            
+            - Best results with clear speech and minimal background noise
+            - Supports various audio formats and durations
+            - Optimized for Bambara language patterns and phonetics
+            """
+        )
+        
+
+        transcribe_btn.click(
+            fn=transcribe_audio,
+            inputs=[audio_input],
+            outputs=output_text,
+            show_progress=True
+        )
+        
+        clear_btn.click(
+            fn=lambda: (None, ""),
+            outputs=[audio_input, output_text]
+        )
+        
+
+        audio_input.change(
+            fn=transcribe_audio,
+            inputs=[audio_input],
+            outputs=output_text,
+            show_progress=True
+        )
+    
+    return demo
+
+def main():
+    """
+    Main function to launch the Gradio interface.
+    """
+    logger.info("Starting Bambara ASR Gradio interface.")
     
+
+    interface = build_interface()
     interface.launch(
         share=False,
         server_name="0.0.0.0",