app.py

import streamlit as st
import requests
import numpy as np
import os
import base64
from streamlit import session_state as st_state
import streamlit_vertical_slider as svs
import librosa
import soundfile as sf
from scipy.signal import butter, sosfilt
 
from wordpress_xmlrpc.compat import xmlrpc_client
from xmlrpc.client import ServerProxy
from io import BytesIO
import soundfile as sf
from datetime import datetime

def save_to_wordpress(audio, sample_rate):
    # Define your WordPress site URL and authentication credentials
    wordpress_url = 'https://songlabai.com/xmlrpc.php'
    username = 'admin_h2ibbgql'
    password = 'um^VdaNK0H8Vw2*KNJlYABkh'

    # Authenticate with WordPress XML-RPC API
    proxy = ServerProxy(wordpress_url)
    wav_bytes = BytesIO()
    sf.write(wav_bytes, audio, samplerate=sample_rate, format='WAV')

    file_data = {
        'name': f"generated_audio_{datetime.now().timestamp()}.wav",
        'type': 'audio/x-wav',  # Change the MIME type according to your file type
        'bits': xmlrpc_client.Binary(wav_bytes.getvalue()),
    }

    for _ in range(4):
        try:
            # Upload the file to WordPress Media Library
            response_upload = proxy.wp.uploadFile(0, username, password, file_data)

            # Handle the response
            if response_upload['id']:
                print(response_upload)
                attachment_id = response_upload['id']
                print("File successfully uploaded to WordPress with attachment ID:", attachment_id)
            else:
                print("Error uploading file to WordPress.")
            break
        except Exception as e:
            print("Error:", e)

# Try to get API_URL from environment variables, if not found set to a default value
try:
    API_URL = os.environ["API_URL"]
except KeyError:
    st.error("API_URL environment variable is not set.")
    st.stop()

# Try to get the Bearer token from environment variables, if not found set to a default value
try:
    BEARER_TOKEN = os.environ["ACCESS_TOKEN"]
except KeyError:
    st.error("BEARER_TOKEN environment variable is not set.")
    st.stop()

headers = {
    "Authorization": f"Bearer {BEARER_TOKEN}",
    "Content-Type": "application/json"
}

# Initialize session state variables
if 'audio' not in st_state:
    st_state.audio = None

if 'augmented_audio' not in st_state:
    st_state.augmented_audio = None

# Streamlit app title
st.title("Songlabai")

# Initialize session state variables
if 'vocal_audio' not in st_state:
    st_state.vocal_audio = None
if 'audio' not in st_state:
    st_state.audio = None
if 'augmented_audio' not in st_state:
    st_state.augmented_audio = None
    
 
genres = [
    "Pop", "Rock", "Hip Hop", "Jazz", "Blues",
    "Country", "Classical", "Electronic", "Reggae",
    "Folk", "R&B", "Metal", "Punk", "Indie",
    "Dance", "World", "Gospel", "Soul", "Funk",
    "Ambient", "Techno", "Disco", "House", "Trance",
    "Dubstep"
]

genre = st.selectbox("Select Genre:", genres)
energy_levels = ["Low", "Medium", "High"]
energy_level = st.radio("Energy Level:", energy_levels, horizontal=True)
description = st.text_input("Description:", "")
tempo = st.slider("Tempo (in bpm):", min_value=40, max_value=100, value=60, step=5)

# Duration input
duration = st.slider("Duration (in seconds):", min_value=15, max_value=90, value=30, step=1)

# Generate audio based on the user's prompt
if st.button("Generate Audio") and genre and energy_level and description and tempo:
    prompt = f"{genre} ,{energy_level}, {tempo}, {description}"
    payload = {"inputs": {"prompt": prompt, "duration": duration}}
    st.text("Generating audio...")
    response = requests.post(API_URL, headers=headers, json=payload)
    response = response.json()[0]
    audio = np.array(response['generated_audio'], dtype=np.float32)
    sample_rate = response['sample_rate']
    st_state.audio = audio
    save_to_wordpress(audio, sample_rate)
    st.audio(st_state.audio, format="audio/wav", sample_rate=sample_rate, start_time=0)

# Post-processing options
st.header("Post-processing Options")

vocal_file = st.file_uploader("Upload Vocal File", type=["mp3", "wav", "ogg", "flac", "aac"])
if vocal_file:
    st_state.vocal_audio = vocal_file.read()
    
# Mixing
mix_vocals = st.checkbox("Mix Vocals")
if mix_vocals and st_state.vocal_audio is not None:
    # Load the vocal audio
    vocal_audio, _ = librosa.load(vocal_file.read(), mono=False)

    # Adjust the vocal audio length to match the generated audio
    vocal_audio = librosa.util.fix_length(vocal_audio, len(st_state.audio))

    # Mix the vocal audio with the generated audio
    st_state.augmented_audio = (st_state.audio + vocal_audio) / 2

# Mastering
st.subheader("Mastering")

# Volume Balance, Compression Ratio, and Reverb Amount
vol_col, comp_col, reverb_col = st.columns(3)

with vol_col:
    volume_balance = svs.vertical_slider(
        "Volume Balance", min_value=-10.0, max_value=10.0, default_value=0.0, step=0.1,
        slider_color="green", track_color="lightgray", thumb_color="red"
    )
    vol_button = st.button("Apply Volume Balance")

with comp_col:
    compression_ratio = svs.vertical_slider(
        "Compression Ratio", min_value=1.0, max_value=10.0, default_value=3.0, step=0.1,
        slider_color="blue", track_color="lightgray", thumb_color="navy"
    )
    comp_button = st.button("Apply Compression")

with reverb_col:
    reverb_amount = svs.vertical_slider(
        "Reverb Amount", min_value=1.0, max_value=10.0, default_value=3.0, step=0.1,
        slider_color="orange", track_color="lightgray", thumb_color="darkorange"
    )
    reverb_button = st.button("Apply Reverb")

if vol_button:
    if st_state.augmented_audio is None:
        st_state.augmented_audio = st_state.audio
    st_state.augmented_audio *= 10 ** (volume_balance / 20)

if comp_button:
    if st_state.augmented_audio is None:
        st_state.augmented_audio = st_state.audio
    # Apply compression using a simple soft-knee compressor
    threshold = -20  # dBFS
    ratio = compression_ratio
    knee = 10  # dB
    max_gain = 20  # dB

    def compress(x, threshold, ratio, knee, max_gain):
        over = np.maximum(x - threshold, 0)
        gain = over / (over + knee) * (1 - (1 / ratio)) + 1
        gain = np.maximum(gain, 1 - max_gain)
        return x * gain

    st_state.augmented_audio = compress(st_state.augmented_audio, threshold, ratio, knee, max_gain)

if reverb_button:
    if st_state.augmented_audio is None:
        st_state.augmented_audio = st_state.audio
    # Apply a simple reverb effect using convolution
    ir_length = int(sample_rate * 2.5)  # 2.5 seconds
    ir = np.zeros(ir_length)
    ir[0] = 1
    ir = np.append(ir, np.zeros(len(st_state.augmented_audio) - ir_length))
    reverb = np.convolve(st_state.augmented_audio, ir, mode='full')[:len(st_state.augmented_audio)]
    st_state.augmented_audio = st_state.augmented_audio + reverb_amount * reverb
    
# Create two main columns
col1, col2 = st.columns(2)

# EQ Controls
with col1:
    eq_cols1 = st.columns(3)
    with eq_cols1[0]:
        eq_low = svs.vertical_slider(
            "Low", default_value=0.0, step=0.1, min_value=-10.0, max_value=10.0,
            slider_color="green", track_color="lightgray", thumb_color="red"
        )
    with eq_cols1[1]:
        eq_mid = svs.vertical_slider(
            "Medium", default_value=0.0, step=0.1, min_value=-10.0, max_value=10.0,
            slider_color="green", track_color="lightgray", thumb_color="red"
        )
    with eq_cols1[2]:
        eq_high = svs.vertical_slider(
            "High", default_value=0.0, step=0.1, min_value=-10.0, max_value=10.0,
            slider_color="green", track_color="lightgray", thumb_color="red"
        )

    eq_cols2 = st.columns(3)
    with eq_cols2[0]:
        pass
    with eq_cols2[1]:
        eq_button = st.button("Apply EQ")
    with eq_cols2[2]:
        pass

# Delay Controls
with col2:
    delay_cols1 = st.columns(2)
    with delay_cols1[0]:
        delay_amount = svs.vertical_slider(
            "Value", default_value=25, step=1, min_value=0, max_value=100,
            slider_color="#6A5ACD", track_color="#F5F5F5", thumb_color="#4B0082"
        )
    with delay_cols1[1]:
        delay_time = svs.vertical_slider(
            "Time", default_value=75, step=1, min_value=0, max_value=100,
            slider_color="#3CB371", track_color="#F5F5F5", thumb_color="#2E8B57"
        )

    delay_cols2 = st.columns(3)
    with delay_cols2[0]:
        pass
    with delay_cols2[1]:
        delay_button = st.button("Delay")
    with delay_cols2[2]:
        pass

if eq_button:
    if st_state.augmented_audio is None:
        st_state.augmented_audio = st_state.audio
    # Apply a simple 3-band EQ using a butterworth filter
    nyquist = sample_rate / 2
    low_cutoff = 200 / nyquist
    mid_cutoff = 2000 / nyquist
    high_cutoff = 8000 / nyquist
    low_sos = butter(4, low_cutoff, btype='low', output='sos', analog=False)
    mid_sos = butter(4, [low_cutoff, mid_cutoff], btype='band', output='sos', analog=False)
    high_sos = butter(4, high_cutoff, btype='high', output='sos', analog=False)
    st_state.augmented_audio = sosfilt(np.dstack((low_sos, mid_sos, high_sos)), st_state.augmented_audio, np.stack((eq_low, eq_mid, eq_high)))

# Add delay logic here if needed
if delay_button:
    if st_state.augmented_audio is None:
        st_state.augmented_audio = st_state.audio
    # Apply a simple delay effect
    delay_samples = int(delay_time / 1000 * sample_rate)
    delay = np.zeros(len(st_state.augmented_audio) + delay_samples)
    delay[delay_samples:] = st_state.augmented_audio
    delay[:len(st_state.augmented_audio)] += delay_amount * delay[:-delay_samples]
    st_state.augmented_audio = delay[:len(st_state.augmented_audio)]

# Display the final audio
if st_state.augmented_audio is not None:
    st.audio(st_state.augmented_audio, format="audio/wav", sample_rate=sample_rate, start_time=0)

# Redirect Button for subscribe
st.link_button("Download/Save", "https://songlabai.com/subcribe/")