Tacotron2 Inference教程

目录结构

本教程实验环境为Google Colab，文件目录结构如下

ALL
└── tacotron2
    ├── audio_processing.py
    ├── checkpoint_269000
    ├── data_utils.py
    ├── demo.wav
    ├── distributed.py
    ├── Dockerfile
    ├── filelists
    │   ├── ljs_audio_text_test_filelist.txt
    │   ├── ljs_audio_text_train_filelist.txt
    │   └── ljs_audio_text_val_filelist.txt
    ├── hparams.py
    ├── inference.ipynb
    ├── layers.py
    ├── LICENSE
    ├── logger.py
    ├── loss_function.py
    ├── loss_scaler.py
    ├── model.py
    ├── multiproc.py
    ├── plotting_utils.py
    ├── __pycache__
    │   ├── audio_processing.cpython-36.pyc
    │   ├── data_utils.cpython-36.pyc
    │   ├── distributed.cpython-36.pyc
    │   ├── hparams.cpython-36.pyc
    │   ├── layers.cpython-36.pyc
    │   ├── logger.cpython-36.pyc
    │   ├── loss_function.cpython-36.pyc
    │   ├── model.cpython-36.pyc
    │   ├── plotting_utils.cpython-36.pyc
    │   ├── stft.cpython-36.pyc
    │   ├── train.cpython-36.pyc
    │   └── utils.cpython-36.pyc
    ├── README.md
    ├── requirements.txt
    ├── stft.py
    ├── tensorboard.png
    ├── text
    │   ├── cleaners.py
    │   ├── cmudict.py
    │   ├── __init__.py
    │   ├── LICENSE
    │   ├── numbers.py
    │   ├── __pycache__
    │   │   ├── cleaners.cpython-36.pyc
    │   │   ├── cmudict.cpython-36.pyc
    │   │   ├── __init__.cpython-36.pyc
    │   │   ├── numbers.cpython-36.pyc
    │   │   └── symbols.cpython-36.pyc
    │   └── symbols.py
    ├── train.py
    ├── utils.py
    └── waveglow
        ├── config.json
        ├── convert_model.py
        ├── denoiser.py
        ├── distributed.py
        ├── glow_old.py
        ├── glow.py
        ├── inference.py
        ├── LICENSE
        ├── mel2samp.py
        ├── __pycache__
        │   ├── denoiser.cpython-36.pyc
        │   └── glow.cpython-36.pyc
        ├── README.md
        ├── requirements.txt
        ├── tacotron2
        ├── train.py
        ├── waveglow_256channels_universal_v5.pt
        └── waveglow_logo.png

文件准备

首先请读者创建一个名为ALL的空文件夹，通过git clone https://github.com/NVIDIA/tacotron2.git命令将tacotron2完整的代码文件下载下来。此时ALL文件夹里面会多出一个名为tacotron2的文件夹，在这个文件夹里有一个inference.ipynb文件，就是等会要用到的推理部分的代码

接着将预训练好的WaveGlow模型保存到waveglow文件夹中（该模型名为waveglow_256channels_universal_v5.pt）

最后还需要一个最重要的文件，就是tacotron2训练时保存的模型文件，一般在训练过程中，它会自动命名为checkpoint_xxxx，将其放到tacotron2文件夹下。如果你自己没有训练tacotron2，官方也提供了一个训练好的模型文件

修改Inference代码

再次强调，我的实验环境是Colab，以下内容均为，文字解释在上，对应代码在下

首先需要确保tensorflow版本为1.x，否则会报错

%tensorflow_version 1.x
import tensorflow as tf
tf.__version__

然后进入ALL/tacotron2目录

%cd ALL/tacotron2

执行代码前需要确保已经安装了unidecode库

!pip install unidecode

导入库，定义函数

import matplotlib
%matplotlib inline
import matplotlib.pylab as plt

import IPython.display as ipd

import sys
sys.path.append('waveglow/')
import numpy as np
import torch

from hparams import create_hparams
from model import Tacotron2
from layers import TacotronSTFT, STFT
from audio_processing import griffin_lim
from train import load_model
from text import text_to_sequence
from denoiser import Denoiser

def plot_data(data, figsize=(16, 4)):
    fig, axes = plt.subplots(1, len(data), figsize=figsize)
    for i in range(len(data)):
        axes[i].imshow(data[i], aspect='auto', origin='bottom', 
                       interpolation='none')
        
hparams = create_hparams()
hparams.sampling_rate = 21050 # 该参数会影响生成语音的语速，越大则语速越快

checkpoint_path = "checkpoint_269000"
model = load_model(hparams)
model.load_state_dict(torch.load(checkpoint_path)['state_dict'])
_ = model.cuda().eval().half()

接着进入waveglow目录加载waveglow模型

%cd ALL/tacotron2/waveglow

waveglow_path = 'waveglow_256channels_universal_v5.pt'
waveglow = torch.load(waveglow_path)['model']
waveglow.cuda().eval().half()
for k in waveglow.convinv:
    k.float()
denoiser = Denoiser(waveglow)

输入文本

text = "WaveGlow is really awesome!"
sequence = np.array(text_to_sequence(text, ['english_cleaners']))[None, :]
sequence = torch.autograd.Variable(
    torch.from_numpy(sequence)).cuda().long()

生成梅尔谱输出，以及画出attention图

mel_outputs, mel_outputs_postnet, _, alignments = model.inference(sequence)
plot_data((mel_outputs.float().data.cpu().numpy()[0],
           mel_outputs_postnet.float().data.cpu().numpy()[0],
           alignments.float().data.cpu().numpy()[0].T))

使用waveglow将梅尔谱合成为语音

with torch.no_grad():
    audio = waveglow.infer(mel_outputs_postnet, sigma=0.666)
ipd.Audio(audio[0].data.cpu().numpy(), rate=hparams.sampling_rate)

（可选）移除waveglow的bias

audio_denoised = denoiser(audio, strength=0.01)[:, 0]
ipd.Audio(audio_denoised.cpu().numpy(), rate=hparams.sampling_rate)