Remove Tensorflow requeriment (#1225)

* Remove TF modules

* Remove TF unit tests

* Remove TF vocoder modules

* Remove TF convert scripts

* Remove TF requirement

* Remove the Docs TF instructions

* Remove TF inference support

tests/vocoder_tests/test_vocoder_tf_melgan_generator.py

@@ -1,19 +0,0 @@
-import unittest
-
-import numpy as np
-import tensorflow as tf
-import torch
-
-from TTS.vocoder.tf.models.melgan_generator import MelganGenerator
-
-use_cuda = torch.cuda.is_available()
-
-
-@unittest.skipIf(use_cuda, " [!] Skip Test: Loosy TF support.")
-def test_melgan_generator():
-    hop_length = 256
-    model = MelganGenerator()
-    # pylint: disable=no-value-for-parameter
-    dummy_input = tf.random.uniform((4, 80, 64))
-    output = model(dummy_input, training=False)
-    assert np.all(output.shape == (4, 1, 64 * hop_length)), output.shape

tests/vocoder_tests/test_vocoder_tf_pqmf.py

@@ -1,31 +0,0 @@
-import os
-import unittest
-
-import soundfile as sf
-import tensorflow as tf
-import torch
-from librosa.core import load
-
-from tests import get_tests_input_path, get_tests_output_path, get_tests_path
-from TTS.vocoder.tf.layers.pqmf import PQMF
-
-TESTS_PATH = get_tests_path()
-WAV_FILE = os.path.join(get_tests_input_path(), "example_1.wav")
-use_cuda = torch.cuda.is_available()
-
-
-@unittest.skipIf(use_cuda, " [!] Skip Test: Loosy TF support.")
-def test_pqmf():
-    w, sr = load(WAV_FILE)
-
-    layer = PQMF(N=4, taps=62, cutoff=0.15, beta=9.0)
-    w, sr = load(WAV_FILE)
-    w2 = tf.convert_to_tensor(w[None, None, :])
-    b2 = layer.analysis(w2)
-    w2_ = layer.synthesis(b2)
-    w2_ = w2.numpy()
-
-    print(w2_.max())
-    print(w2_.min())
-    print(w2_.mean())
-    sf.write(os.path.join(get_tests_output_path(), "tf_pqmf_output.wav"), w2_.flatten(), sr)