app.py

import random 
import gradio as gr 
import numpy as np 
from elevenlabs import voices, generate, set_api_key, UnauthenticatedRateLimitError

def pad_buffer(audio):
    # Pad buffer to multiple of 2 bytes
    buffer_size = len(audio)
    element_size = np.dtype(np.int16).itemsize
    if buffer_size % element_size != 0:
        audio = audio + b'\0' * (element_size - (buffer_size % element_size))
    return audio 

def generate_voice(text, voice_name):
    try:
        audio = generate(
            text[:4000], # Limit to 250 characters
            voice=voice_name, 
            model="eleven_multilingual_v2"
        )
        return (44100, np.frombuffer(pad_buffer(audio), dtype=np.int16))
    except UnauthenticatedRateLimitError as e:
        raise gr.Error("Thanks for trying out ElevenLabs TTS! You've reached the free tier limit. Please provide an API key to continue.") 
    except Exception as e:
        raise gr.Error(e)
    

badges = """
<div style="display: flex">
<span style="margin-right: 5px"> 

[ ![GitHub](https://img.shields.io/badge/github-%23121011.svg?style=for-the-badge&logo=github&logoColor=white) ](https://github.com/elevenlabs/elevenlabs-python)
 
</span>
<span style="margin-right: 5px"> 

[ ![Twitter](https://img.shields.io/badge/Twitter-%231DA1F2.svg?style=for-the-badge&logo=Twitter&logoColor=white) ](https://twitter.com/elevenlabsio)
 
</span>
<span>

[ ![](https://dcbadge.vercel.app/api/server/elevenlabs) ](https://discord.gg/elevenlabs)

</span>
</div>
"""

description = """
A demo of the world's most advanced TTS systems, made by [ElevenLabs](https://elevenlabs.io). Eleven Multilingual V2 is a single foundational model supporting 28 languages including: English, Chinese, Spanish, Hindi, Portuguese, French, German, Japanese, Arabic, Korean, Indonesian, Italian, Dutch, Turkish, Polish, Swedish, Filipino, Malay, Romanian, Ukrainian, Greek, Czech, Danish, Finnish, Bulgarian, Croatian, Slovak, and Tamil. Sign up on [ElevenLabs](https://elevenlabs.io) to get fast access, long-form generation, voice cloning, API keys, and more!
"""

with gr.Blocks() as block:
    gr.Markdown('[ ![ElevenLabs](https://user-images.githubusercontent.com/12028621/262629275-4f85c9cf-85b6-435e-ab50-5b8c7c4e9dd2.png) ](https://elevenlabs.io)')
    gr.Markdown(badges)
    gr.Markdown(description)
    
    input_text = gr.Textbox(
        label="Input Text (4000 characters max)", 
        lines=2, 
        value="Hello! 你好! Hola! नमस्ते! Bonjour! こんにちは! مرحبا! 안녕하세요! Ciao! Cześć! Привіт! Γειά σας! Здравей! வணக்கம்!",
        elem_id="input_text"
    )

    all_voices = voices() 
    input_voice = gr.Dropdown(
        [ voice.name for voice in all_voices ], 
        value="Bella",
        label="Voice", 
        elem_id="input_voice"
    )

    run_button = gr.Button(
        text="Generate Voice", 
        type="button"
    )

    out_audio = gr.Audio(
        label="Generated Voice",
        type="numpy", 
        elem_id="out_audio",
        format="mp3"
    )
        
    inputs = [input_text, input_voice]
    outputs = [out_audio]
    
    run_button.click(
        fn=generate_voice, 
        inputs=inputs, 
        outputs=outputs, 
        queue=True
    )

block.queue(concurrency_count=5).launch(debug=True)