File size: 7,123 Bytes
8dae870 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 | import gradio as gr
import os
import soundfile as sf
from core.cloner import KokoClone
# 1. Initialize the cloner globally so models load only once when the server starts
print("Loading KokoClone models for the Web UI...")
cloner = KokoClone()
def clone_voice(text, lang, ref_audio_path):
"""Gradio handler: text + reference audio → cloned speech."""
if not text or not text.strip():
raise gr.Error("Please enter some text.")
if not ref_audio_path:
raise gr.Error("Please upload or record a reference audio file.")
output_file = "gradio_output.wav"
try:
cloner.generate(
text=text,
lang=lang,
reference_audio=ref_audio_path,
output_path=output_file
)
return output_file
except Exception as e:
raise gr.Error(f"An error occurred during generation: {str(e)}")
def convert_voice(source_audio_path, ref_audio_path):
"""Gradio handler: source audio + reference audio → re-voiced speech."""
if not source_audio_path:
raise gr.Error("Please upload or record a source audio file.")
if not ref_audio_path:
raise gr.Error("Please upload or record a reference audio file.")
output_file = "gradio_convert_output.wav"
try:
cloner.convert(
source_audio=source_audio_path,
reference_audio=ref_audio_path,
output_path=output_file
)
return output_file
except Exception as e:
raise gr.Error(f"An error occurred during conversion: {str(e)}")
# 2. Build the Gradio UI using Blocks
with gr.Blocks() as demo:
gr.Markdown(
"""
<div style="text-align: center;">
<h1>🎧 KokoClone</h1>
<p>Voice Cloning, Now Inside Kokoro.<br>
Generate natural multilingual speech and clone any target voice with ease.<br>
<i>Built on Kokoro TTS.</i></p>
</div>
"""
)
with gr.Tabs():
# ── Tab 1: Text → Cloned Speech ─────────────────────────────────────
with gr.Tab("🎤 Text → Clone"):
with gr.Row():
with gr.Column(scale=1):
text_input = gr.Textbox(
label="1. Text to Synthesize",
lines=4,
placeholder="Enter the text you want spoken..."
)
lang_input = gr.Dropdown(
label="2. Language",
choices=[
("English", "en"),
("Hindi", "hi"),
("French", "fr"),
("Japanese", "ja"),
("Chinese", "zh"),
("Italian", "it"),
("Spanish", "es"),
("Portuguese", "pt")
],
value="en"
)
ref_audio_input = gr.Audio(
label="3. Reference Voice (Upload or Record)",
type="filepath"
)
submit_btn = gr.Button("🚀 Generate Clone", variant="primary")
with gr.Column(scale=1):
output_audio = gr.Audio(
label="Generated Cloned Audio",
interactive=False,
autoplay=False
)
gr.Markdown(
"""
<br>
### 💡 Tips for Best Results:
* **Clean Audio:** Use a reference audio clip without background noise or music.
* **Length:** A reference clip of 3 to 10 seconds is usually the sweet spot.
* **Language Match:** Make sure the selected language matches the text you typed!
* **First Run:** The very first generation might take a few extra seconds while the models allocate memory.
"""
)
submit_btn.click(
fn=lambda: gr.update(value="⌛ Generating...", interactive=False),
outputs=submit_btn
).then(
fn=clone_voice,
inputs=[text_input, lang_input, ref_audio_input],
outputs=output_audio
).then(
fn=lambda: gr.update(value="🚀 Generate Clone", interactive=True),
outputs=submit_btn
)
# ── Tab 2: Audio → Re-voiced Speech ─────────────────────────────────
with gr.Tab("🔁 Audio → Clone"):
with gr.Row():
with gr.Column(scale=1):
source_audio_input = gr.Audio(
label="1. Source Audio (speech to re-voice)",
type="filepath"
)
ref_audio_convert_input = gr.Audio(
label="2. Reference Voice (target speaker)",
type="filepath"
)
convert_btn = gr.Button("🔁 Convert Voice", variant="primary")
with gr.Column(scale=1):
convert_output_audio = gr.Audio(
label="Converted Audio",
interactive=False,
autoplay=False
)
gr.Markdown(
"""
<br>
### 💡 How it works:
* Upload any speech recording as the **source**.
* Upload a short clip of the **target speaker** as the reference.
* KokoClone re-voices the source speech to sound like the reference — no transcription needed.
### Tips:
* Clean, noise-free audio works best for both inputs.
* Reference clips of 3–10 seconds give the best voice transfer.
"""
)
convert_btn.click(
fn=lambda: gr.update(value="⌛ Converting...", interactive=False),
outputs=convert_btn
).then(
fn=convert_voice,
inputs=[source_audio_input, ref_audio_convert_input],
outputs=convert_output_audio
).then(
fn=lambda: gr.update(value="🔁 Convert Voice", interactive=True),
outputs=convert_btn
)
# 4. Launch the app
if __name__ == "__main__":
# Gradio 6.0 fix: Moved theme here and removed show_api
demo.launch(server_name="0.0.0.0")
|