openvoice_app.py

import os
import torch
import argparse
import gradio as gr
import requests
from openvoice import se_extractor
from openvoice.api import BaseSpeakerTTS, ToneColorConverter
from dotenv import load_dotenv
from openai import OpenAI
from elevenlabs.client import ElevenLabs
from elevenlabs import play, save
from flask import Flask
from flask_limiter import Limiter
from flask_limiter.util import get_remote_address

# Load environment variables
load_dotenv()

# Initialize Flask app
app = Flask(__name__)

# Setup Limiter for rate limiting and quota management based on IP address
limiter = Limiter(get_remote_address, app=app, default_limits=["5 per minute"])

# Argument parsing
parser = argparse.ArgumentParser()
parser.add_argument("--share", action='store_true', default=False, help="make link public")
args = parser.parse_args()

# Initialize ElevenLabs client
client = ElevenLabs(api_key=os.environ.get("ELEVENLABS_API_KEY"))
device = 'cuda' if torch.cuda.is_available() else 'cpu'
output_dir = 'outputs'
os.makedirs(output_dir, exist_ok=True)

api_key = os.environ.get("ELEVENLABS_API_KEY")
supported_languages = ['zh', 'en']

# Function to get all voices
def get_voices(api_key):
    url = "https://api.elevenlabs.io/v1/voices"
    headers = {"xi-api-key": api_key}
    response = requests.request("GET", url, headers=headers)
    return response.json()

# Function to delete a voice by ID
def delete_voice(api_key, voice_id):
    url = f"https://api.elevenlabs.io/v1/voices/{voice_id}"
    headers = {"xi-api-key": api_key}
    response = requests.request("DELETE", url, headers=headers)
    return response.status_code, response.text

# Predict function with rate limiting based on IP address
#@limiter.limit("100 per minute")
def predict(prompt, style, audio_file_pth, voice_name):
    text_hint = ''
    if len(prompt) < 2:
        text_hint += "[ERROR] Please provide a longer prompt text.\n"
        return text_hint, None, None
    if len(prompt) > 200:
        text_hint += "[ERROR] Text length limited to 200 characters. Please try shorter text.\n"
        return text_hint, None, None
    
    print(audio_file_pth)
    voice = client.clone(
        name=voice_name,
        description="A trial voice model for testing",
        files=[audio_file_pth],
    )
    # Generate audio from text
    audio = client.generate(text=prompt, voice=voice)
    save(audio, f'{output_dir}/output.wav')
    
    save_path = f'{output_dir}/output.wav'
    data = get_voices(api_key)
    # Find all voice IDs with the name provided by the user
    trial_voice_ids = [voice.get("voice_id") for voice in data['voices'] if voice.get("name") == voice_name]

    # # Delete each voice with the name provided by the user
    # for voice_id in trial_voice_ids:
    #     status_code, response_text = delete_voice(api_key, voice_id)
    #     print(f"Deleted voice ID {voice_id}: Status Code {status_code}, Response {response_text}")

    # if not trial_voice_ids:
    #     print("No voices with the name provided by the user found.")

    return text_hint, save_path, audio_file_pth

# Gradio interface setup
with gr.Blocks(gr.themes.Glass()) as demo:
    with gr.Row():
        with gr.Column():
            input_text_gr = gr.Textbox(
                label="Text Prompt",
                info="One or two sentences at a time is better. Up to 200 text characters.",
                value="He hoped there would be stew for dinner, turnips and carrots and bruised potatoes and fat mutton pieces to be ladled out in thick, peppered, flour-fattened sauce.",
            )
            style_gr = gr.Dropdown(
                label="Style",
                choices=['default', 'whispering', 'cheerful', 'terrified', 'angry', 'sad', 'friendly'],
                info="Please upload a reference audio file that is at least 1 minute long. For best results, ensure the audio is clear. You can use Adobe Podcast Enhance(https://podcast.adobe.com/enhance) to improve the audio quality before uploading.",
                max_choices=1,
                value="default",
            )
            ref_gr = gr.Audio(
                label="Reference Audio",
                type="filepath",
                value="resources/demo_speaker2.mp3",
                sources=["upload"],  # Allow only upload
            )
            voice_name_gr = gr.Textbox(
                label="Voice Name",
                info="Name for the cloned voice model.",
                value="TrialVoice"
            )
            tts_button = gr.Button("Send", elem_id="send-btn", visible=True)

        with gr.Column():
            out_text_gr = gr.Text(label="Info")
            audio_gr = gr.Audio(label="Synthesised Audio", autoplay=True)
            ref_audio_gr = gr.Audio(label="Reference Audio Used")

            tts_button.click(predict, [input_text_gr, style_gr, ref_gr, voice_name_gr], outputs=[out_text_gr, audio_gr, ref_audio_gr])

    demo.queue()
    demo.launch(debug=True, show_api=False, share=args.share)

# Hide Gradio footer and record button
css = """
footer {visibility: hidden}
audio .btn-container {display: none}
"""

demo.add_css(css)