app.py

import os



from transformers import BarkModel
from transformers import AutoProcessor

###### BARK
model_bark = BarkModel.from_pretrained("suno/bark")
processor_bark = AutoProcessor.from_pretrained("suno/bark")
###### FACEBOOK
#from mms import generate_audio_mms
from tts import synthesize, TTS_LANGUAGES
import torch

device = "cuda:0" if torch.cuda.is_available() else "cpu"
model_bark = model_bark.to(device)

def generate_audio_bark(input):
  voice_preset = "v2/es_speaker_1"

  inputs = processor_bark(input, voice_preset=voice_preset)
  # generate speech
  sampling_rate = model_bark.generation_config.sample_rate
  speech_output = model_bark.generate(**inputs.to(device))
  return sampling_rate,speech_output[0].cpu().numpy()


import gradio as gr
with gr.Blocks() as demo:

  with gr.Tab("MMS"):
    with gr.Row():
      with gr.Column():
        textbox = gr.Textbox(label="Ingrese texto")
        button = gr.Button("Hablar")
      with gr.Column():
        audio_output = gr.Audio()
  with gr.Tab("Bark"):
    with gr.Row():
      with gr.Column():
        textbox2= gr.Textbox(label="Ingrese text")
        button2 = gr.Button("Hablar")
      with gr.Column():
        audio_output2 = gr.Audio()


    button.click(synthesize,textbox,audio_output)
    button2.click(generate_audio_bark,textbox2,audio_output2)

demo.launch()