import torch
import torchaudio
import gradio as gr

device="cpu"
bundle = torchaudio.pipelines.TACOTRON2_WAVERNN_PHONE_LJSPEECH
processor = bundle.get_text_processor()
tacotron2 = bundle.get_tacotron2().to(device)

# Workaround to load model mapped on GPU
# https://stackoverflow.com/a/61840832
waveglow = torch.hub.load(
    "NVIDIA/DeepLearningExamples:torchhub",
    "nvidia_waveglow",
    model_math="fp32",
    pretrained=False,
)
checkpoint = torch.hub.load_state_dict_from_url(
    "https://api.ngc.nvidia.com/v2/models/nvidia/waveglowpyt_fp32/versions/1/files/nvidia_waveglowpyt_fp32_20190306.pth",  # noqa: E501
    progress=False,
    map_location=device,
)
state_dict = {key.replace("module.", ""): value for key, value in checkpoint["state_dict"].items()}

waveglow.load_state_dict(state_dict)
waveglow = waveglow.remove_weightnorm(waveglow)
waveglow = waveglow.to(device)
waveglow.eval()

def inference(text):

  with torch.inference_mode():
      processed, lengths = processor(text)
      processed = processed.to(device)
      lengths = lengths.to(device)
      spec, _, _ = tacotron2.infer(processed, lengths)



  with torch.no_grad():
      waveforms = waveglow.infer(spec)

  torchaudio.save("output_waveglow.wav", waveforms[0:1].cpu(), sample_rate=22050)
  return "output_waveglow.wav"
  
title="WAVEGLOW"
description="Gradio demo for WAVEGLOW: WaveGlow model for generating speech from mel spectrograms (generated by Tacotron2). To use it, simply add you text or click on one of the examples to load them. Read more at the links below."

article = "<p style='text-align: center'><a href='https://arxiv.org/abs/1811.00002' target='_blank'>WaveGlow: A Flow-based Generative Network for Speech Synthesis</a> | <a href='https://github.com/NVIDIA/DeepLearningExamples/tree/master/PyTorch/SpeechSynthesis/Tacotron2' target='_blank'>Github Repo</a></p>"
examples=[["life is like a box of chocolates"]]
gr.Interface(inference,"text",gr.outputs.Audio(type="file"),title=title,description=description,article=article,examples=examples).launch(enable_queue=True)