app.py

import gradio as gr
import matplotlib.pyplot as plt
from matplotlib.colors import LinearSegmentedColormap
import numpy as np
import os
import soundfile as sf
import requests

def download_file(url):
    file_id = url.split('/')[-2]
    download_url = f'https://docs.google.com/uc?export=download&id={file_id}'
    response = requests.get(download_url, allow_redirects=True)
    local_filename = url.split('/')[-1] + '.wav'
    open(local_filename, 'wb').write(response.content)
    return local_filename

def main():
    with gr.Blocks() as app:
        gr.Markdown(
            """
            Audio Analyzer Software by Ilaria, Help me on Ko-Fi!\n
            Special thanks to Alex Murkoff for helping me coding it!
            Need help with AI? Join Join AI Hub!
            """
        )
    
        with gr.Row():
            with gr.Column():
                audio_input = gr.Audio(type='filepath')
                create_spec_butt = gr.Button(value='Create Spectrogram And Get Info', variant='primary')

            with gr.Column():
                output_markdown = gr.Markdown(value="", visible=True)
                image_output = gr.Image(type='filepath', interactive=False)
            
                with gr.Accordion('Audio Downloader', open=False):
                    url_input = gr.Textbox(value='', label='Google Drive Audio URL')
                    download_butt = gr.Button(value='Download audio', variant='primary')
                
                download_butt.click(fn=download_file, inputs=[url_input], outputs=[audio_input])
                create_spec_butt.click(fn=create_spectrogram_and_get_info, inputs=[audio_input], outputs=[output_markdown, image_output])
            
        download_butt.click(fn=download_file, inputs=[url_input], outputs=[audio_input])
        create_spec_butt.click(fn=create_spectrogram_and_get_info, inputs=[audio_input], outputs=[output_markdown, image_output])
        
        app.queue(max_size=1022).launch(share=True)

def create_spectrogram_and_get_info(audio_file):
    plt.clf()
    
    cdict = {'red':   [(0.0,  0.5, 0.5),
                       (0.5,  1.0, 1.0),
                       (1.0,  1.0, 1.0)],
             'green': [(0.0,  0.0, 0.0),
                       (0.25, 0.0, 0.0),
                       (0.75, 1.0, 1.0),
                       (1.0,  1.0, 1.0)],
             'blue':  [(0.0,  0.5, 0.5),
                       (0.5,  0.0, 0.0),
                       (1.0,  0.0, 0.0)]}
    custom_cmap = LinearSegmentedColormap('CustomMap', cdict)
    
    plt.figure(figsize=(15, 5))
    audio_data, sample_rate = sf.read(audio_file)
    if len(audio_data.shape) > 1:
        audio_data = np.mean(audio_data, axis=1)
    plt.specgram(audio_data, Fs=sample_rate / 1, NFFT=4096, sides='onesided',
                 cmap=custom_cmap, scale_by_freq=True, scale='dB', mode='magnitude', window=np.hanning(4096))  # Usa la mappa di colori personalizzata
    plt.savefig('spectrogram.png', dpi=300)
    audio_info = sf.info(audio_file)
    bit_depth = {'PCM_16': 16, 'FLOAT': 32}.get(audio_info.subtype, 0)
    minutes, seconds = divmod(audio_info.duration, 60)
    seconds, milliseconds = divmod(seconds, 1)
    milliseconds *= 1000
    bitrate = audio_info.samplerate * audio_info.channels * bit_depth / 8 / 1024 / 1024
    speed_in_kbps = audio_info.samplerate * bit_depth / 1000
    filename_without_extension, _ = os.path.splitext(os.path.basename(audio_file))
    info_table = f"""
    
    | Information | Value |
    | :---: | :---: |
    | File Name | {filename_without_extension} |
    | Duration | {int(minutes)} minutes - {int(seconds)} seconds - {int(milliseconds)} milliseconds |
    | Bitrate | {speed_in_kbps} kbp/s |
    | Audio Channels | {audio_info.channels} |
    | Samples per second | {audio_info.samplerate} Hz |
    | Bit per second | {audio_info.samplerate * audio_info.channels * bit_depth} bit/s |
    
    """
    
    # Return the PNG file of the spectrogram and the info table
    return info_table, 'spectrogram.png'

# Create the Gradio interface
main()