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, api_key):
    set_api_key(api_key) #set API key
    try:
        audio = generate(
            text[:4000], # Limit to 4000 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)
    
description = """
Eleven Multilingual V2 is the world's best Text-to-Speech model. Features 38 voices and supports 28 languages. Sign up on [ElevenLabs](https://elevenlabs.io/?from=partnerpierce7156) to get an API Key.
"""

with gr.Blocks() as block:
    gr.Markdown("# <center> ElevenLabs TTS </center>")
    gr.Markdown(description)

    all_voices = voices() 

with gr.Row(variant='panel'):
    input_api_key = gr.Textbox(
        type='password', 
        label='Elevenlabs API Key', 
        placeholder='Enter your API key',
        elem_id="input_api_key"
    )

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

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"
)

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, input_api_key]
outputs = [out_audio]  # Added missing assignment


run_button.click(
    fn=generate_voice, 
    inputs=inputs, 
    outputs=outputs, 
    queue=True
)

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