app.py

import gradio as gr
from logic import compare_audio_with_text
from scipy.io import wavfile



def create_html_from_scores(word_scores):
    html_output = ''
    
    # Ensure the number of words and scores match
    
    for word, score in word_scores:
        if score == 1:
            html_output += f'<span style="color: red;">{word}</span> '
        elif score == 2:
            html_output += f'<span style="color: orange;">{word}</span> '
        else:
            html_output += f'<span style="color: green;">{word}</span> '
    return html_output




def analyze_audio(text, audio):
# Write the processed audio to a temporary WAV file
    temp_filename = 'temp_audio.wav'
    wavfile.write(temp_filename, audio[0], audio[1])


    result = compare_audio_with_text(temp_filename, text)
    html_content = create_html_from_scores(result)
    html_with_css = f"""
    <style>
    .legend {{
      font-size: 22px;
      display: flex;
      align-items: center;
      gap: 12px;
    }}
    
    .legend-dot {{
      height: 15px;
      width: 15px;
      border-radius: 50%;
      display: inline-block;
    }}
    
    .good {{ color: #28a745; }}
    .average {{ color: #ffc107; }}
    .bad {{ color: #dc3545; }}
    
    .text {{ font-size: 20px; }}
    </style>
    
    <div class="legend">
      <span class="legend-dot" style="background-color: #28a745;"></span><span>Good</span>
      <span class="legend-dot" style="background-color: #ffc107;"></span><span>Understandable</span>
      <span class="legend-dot" style="background-color: #dc3545;"></span><span>Bad</span>
    </div>
    
    <p class="text">
      {html_content}
    </p>
    """
    return html_with_css

# Define the Gradio interface
iface = gr.Interface(fn=analyze_audio,
                     inputs=[gr.Textbox(label='Training Text', placeholder='Write the text for pronunciation task', interactive=True, visible=True, show_copy_button=True,), 
                             gr.Audio(label="Recoreded Audio")
                             ],
                     outputs=[gr.HTML(label="Analysis of pronunciation"),
                              ],
                     # css=additional_css,
                     # title="Audio Analysis Tool",
                     description="Write any text and recored an audio to predict pronunciation erors"
                     )

# Run the Gradio app
if __name__ == "__main__":
    iface.launch()