File size: 3,248 Bytes
88f7073 3cdb410 11023cf e37c63c 3cdb410 e37c63c 471fe68 e37c63c 471fe68 1a0b3dd 471fe68 e37c63c 471fe68 88f7073 471fe68 88f7073 471fe68 88f7073 e37c63c |
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 |
import gradio as gr
import torch
from TTS.api import TTS
import os
import spaces
import tempfile
os.environ["COQUI_TOS_AGREED"] = "1"
device = "cuda" if torch.cuda.is_available() else "cpu"
# Initialize TTS model
def load_tts_model():
return TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
tts = load_tts_model()
# Celebrity voices (example list, you may want to expand or modify this)
celebrity_voices = {
"Morgan Freeman": "path/to/morgan_freeman_sample.wav",
"Scarlett Johansson": "path/to/scarlett_johansson_sample.wav",
"David Attenborough": "path/to/david_attenborough_sample.wav",
}
def tts_generate(text, voice, language):
with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio:
temp_audio_path = temp_audio.name
tts.tts_to_file(
text=text,
speaker_wav=celebrity_voices[voice],
language=language,
file_path=temp_audio_path
)
return temp_audio_path
@spaces.GPU(enable_queue=True)
def clone_voice(text, audio_file, language):
with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio:
temp_audio_path = temp_audio.name
tts.tts_to_file(
text=text,
speaker_wav=audio_file,
language=language,
file_path=temp_audio_path
)
return temp_audio_path
# Placeholder function for Talking Image tab
def talking_image_placeholder():
return "Talking Image functionality not implemented yet."
# Define Gradio interface
with gr.Blocks() as demo:
gr.Markdown("# Advanced Voice Synthesis")
with gr.Tabs():
with gr.TabItem("TTS"):
with gr.Row():
tts_text = gr.Textbox(label="Text to speak")
tts_voice = gr.Dropdown(choices=list(celebrity_voices.keys()), label="Celebrity Voice")
tts_language = gr.Dropdown(["en", "es", "fr", "de", "it"], label="Language", value="en")
tts_generate_btn = gr.Button("Generate")
tts_output = gr.Audio(label="Generated Audio")
tts_generate_btn.click(
tts_generate,
inputs=[tts_text, tts_voice, tts_language],
outputs=tts_output
)
with gr.TabItem("Talking Image"):
gr.Markdown("Talking Image functionality coming soon!")
with gr.TabItem("Clone Voice"):
with gr.Row():
clone_text = gr.Textbox(label="Text to speak")
clone_audio = gr.Audio(label="Voice reference audio file", type="filepath")
clone_language = gr.Dropdown(["en", "es", "fr", "de", "it"], label="Language", value="en")
clone_generate_btn = gr.Button("Generate")
clone_output = gr.Audio(label="Generated Audio")
clone_generate_btn.click(
clone_voice,
inputs=[clone_text, clone_audio, clone_language],
outputs=clone_output
)
# Launch the interface
demo.launch()
# Clean up temporary files (this will run after the Gradio server is closed)
for file in os.listdir():
if file.endswith('.wav') and file.startswith('tmp'):
os.remove(file) |