File size: 5,582 Bytes
e86aa69 7c84ef7 e86aa69 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 |
import gradio as gr
import numpy as np
import librosa
import soundfile as sf
from TTS.api import TTS
import torch
import os
import tempfile
# Initialize TTS model
try:
tts = TTS("tts_models/multilingual/multi-dataset/your_tts", progress_bar=False)
except Exception as e:
print(f"Error initializing TTS model: {e}")
tts = None
def load_audio(audio_path):
try:
audio, sr = librosa.load(audio_path, sr=None)
return audio, sr
except Exception as e:
print(f"Error loading audio: {e}")
return None, None
def save_audio(audio, sr, path):
try:
sf.write(path, audio, sr)
except Exception as e:
print(f"Error saving audio: {e}")
def pitch_shift(audio, sr, n_steps):
try:
return librosa.effects.pitch_shift(audio, sr=sr, n_steps=n_steps)
except Exception as e:
print(f"Error in pitch shifting: {e}")
return audio
def change_voice(audio_path, pitch_shift_amount, formant_shift_amount):
if tts is None:
return None, None
audio, sr = load_audio(audio_path)
if audio is None or sr is None:
return None, None
pitched_audio = pitch_shift(audio, sr, pitch_shift_amount)
try:
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as temp_file:
save_audio(pitched_audio, sr, temp_file.name)
converted_audio_path = tts.voice_conversion(
source_wav=temp_file.name,
target_wav="path/to/female_target_voice.wav", # You need to provide a female target voice file
output_wav=None
)
converted_audio, _ = load_audio(converted_audio_path)
formant_shifted_audio = librosa.effects.pitch_shift(converted_audio, sr=sr, n_steps=formant_shift_amount)
os.unlink(temp_file.name)
os.unlink(converted_audio_path)
return sr, formant_shifted_audio
except Exception as e:
print(f"Error in voice conversion: {e}")
return None, None
def process_audio(audio_file, pitch_shift_amount, formant_shift_amount):
if audio_file is None:
return None
# Use the audio_file path directly
sr, audio = change_voice(audio_file, pitch_shift_amount, formant_shift_amount)
if sr is None or audio is None:
return None
output_path = "output_voice.wav"
save_audio(audio, sr, output_path)
return output_path
# Custom CSS for improved design
custom_css = """
.gradio-container {
background-color: #f0f4f8;
}
.container {
max-width: 900px;
margin: auto;
padding: 20px;
border-radius: 10px;
background-color: white;
box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1);
}
h1 {
color: #2c3e50;
text-align: center;
font-size: 2.5em;
margin-bottom: 20px;
}
.description {
text-align: center;
color: #34495e;
margin-bottom: 30px;
}
.input-section, .output-section {
background-color: #ecf0f1;
padding: 20px;
border-radius: 8px;
margin-bottom: 20px;
}
.input-section h3, .output-section h3 {
color: #2980b9;
margin-bottom: 15px;
}
"""
# Gradio Interface with improved design
with gr.Blocks(css=custom_css) as demo:
gr.HTML(
"""
<div style="text-align: center; max-width: 800px; margin: 0 auto;">
<div style="display: inline-flex; align-items: center; gap: 0.8rem; font-size: 1.75rem;">
<svg xmlns="http://www.w3.org/2000/svg" width="1em" height="1em" fill="currentColor" viewBox="0 0 16 16" style="vertical-align: middle;">
<path d="M3.5 6.5A.5.5 0 0 1 4 7v1a4 4 0 0 0 8 0V7a.5.5 0 0 1 1 0v1a5 5 0 0 1-4.5 4.975V15h3a.5.5 0 0 1 0 1h-7a.5.5 0 0 1 0-1h3v-2.025A5 5 0 0 1 3 8V7a.5.5 0 0 1 .5-.5z"/>
<path d="M10 8a2 2 0 1 1-4 0V3a2 2 0 1 1 4 0v5zM8 0a3 3 0 0 0-3 3v5a3 3 0 0 0 6 0V3a3 3 0 0 0-3-3z"/>
</svg>
<h1 style="font-weight: 900; margin-bottom: 7px;">
AI Voice Changer
</h1>
</div>
<p class="description">Transform any voice into a realistic female voice using advanced AI technology</p>
</div>
"""
)
with gr.Row():
with gr.Column(elem_classes="input-section"):
gr.Markdown("### Input")
audio_input = gr.Audio(type="filepath", label="Upload Voice")
pitch_shift = gr.Slider(-12, 12, step=0.5, label="Pitch Shift", value=0)
formant_shift = gr.Slider(-5, 5, step=0.1, label="Formant Shift", value=0)
submit_btn = gr.Button("Transform Voice", variant="primary")
with gr.Column(elem_classes="output-section"):
gr.Markdown("### Output")
audio_output = gr.Audio(label="Transformed Voice")
submit_btn.click(
fn=process_audio,
inputs=[audio_input, pitch_shift, formant_shift],
outputs=audio_output,
)
gr.Markdown(
"""
### How to use:
1. Upload an audio file containing the voice you want to transform.
2. Adjust the Pitch Shift and Formant Shift sliders to fine-tune the voice (optional).
3. Click the "Transform Voice" button to process the audio.
4. Listen to the transformed voice in the output section.
5. Download the transformed audio file if desired.
Note: This application uses AI to transform voices. The quality of the output may vary depending on the input audio quality and the chosen settings.
"""
)
if __name__ == "__main__":
demo.launch() |