import os
import unittest
from tests import get_tests_input_path, get_tests_output_path, get_tests_path
from TTS.config import BaseAudioConfig
from TTS.utils.audio import AudioProcessor
TESTS_PATH = get_tests_path()
OUT_PATH = os.path.join(get_tests_output_path(), "audio_tests")
WAV_FILE = os.path.join(get_tests_input_path(), "example_1.wav")
os.makedirs(OUT_PATH, exist_ok=True)
conf = BaseAudioConfig(mel_fmax=8000)
# pylint: disable=protected-access
class TestAudio(unittest.TestCase):
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.ap = AudioProcessor(**conf)
def test_audio_synthesis(self):
"""1. load wav
2. set normalization parameters
3. extract mel-spec
4. invert to wav and save the output
"""
print(" > Sanity check for the process wav -> mel -> wav")
def _test(max_norm, signal_norm, symmetric_norm, clip_norm):
self.ap.max_norm = max_norm
meta_file_train="metadata.csv",
path=os.path.join(output_path,
"../thorsten-de/"))
# download dataset if not already present
if not os.path.exists(dataset_config.path):
print("Downloading dataset")
download_thorsten_de(
os.path.split(os.path.abspath(dataset_config.path))[0])
audio_config = BaseAudioConfig(
sample_rate=22050,
do_trim_silence=True,
trim_db=60.0,
signal_norm=False,
mel_fmin=0.0,
mel_fmax=8000,
spec_gain=1.0,
log_func="np.log",
ref_level_db=20,
preemphasis=0.0,
)
config = SpeedySpeechConfig(
run_name="speedy_speech_thorsten-de",
audio=audio_config,
batch_size=32,
eval_batch_size=16,
num_loader_workers=4,
num_eval_loader_workers=4,
compute_input_seq_cache=True,
run_eval=True,
import torch
from tests import get_tests_input_path, get_tests_output_path, get_tests_path
from TTS.config import BaseAudioConfig
from TTS.utils.audio import AudioProcessor
from TTS.vocoder.layers.losses import MelganFeatureLoss, MultiScaleSTFTLoss, STFTLoss, TorchSTFT
TESTS_PATH = get_tests_path()
OUT_PATH = os.path.join(get_tests_output_path(), "audio_tests")
os.makedirs(OUT_PATH, exist_ok=True)
WAV_FILE = os.path.join(get_tests_input_path(), "example_1.wav")
ap = AudioProcessor(**BaseAudioConfig().to_dict())
def test_torch_stft():
torch_stft = TorchSTFT(ap.fft_size, ap.hop_length, ap.win_length)
# librosa stft
wav = ap.load_wav(WAV_FILE)
M_librosa = abs(ap._stft(wav)) # pylint: disable=protected-access
# torch stft
wav = torch.from_numpy(wav[None, :]).float()
M_torch = torch_stft(wav)
# check the difference b/w librosa and torch outputs
assert (M_librosa - M_torch[0].data.numpy()).max() < 1e-5
def test_stft_loss():