diff options
| -rw-r--r-- | become_yukarin/data_struct.py | 2 | ||||
| -rw-r--r-- | scripts/process.py | 7 |
2 files changed, 6 insertions, 3 deletions
diff --git a/become_yukarin/data_struct.py b/become_yukarin/data_struct.py index 6597d0d..4474331 100644 --- a/become_yukarin/data_struct.py +++ b/become_yukarin/data_struct.py @@ -43,7 +43,7 @@ class AcousticFeature(NamedTuple): spectrogram=self.spectrogram.astype(dtype), aperiodicity=self.aperiodicity.astype(dtype), mfcc=self.mfcc.astype(dtype), - voiced=self.voiced.astype(dtype), + voiced=self.voiced, ) def validate(self): diff --git a/scripts/process.py b/scripts/process.py index 31da6aa..8f3473c 100644 --- a/scripts/process.py +++ b/scripts/process.py @@ -10,6 +10,7 @@ import numpy from become_yukarin import AcousticConverter from become_yukarin import SuperResolution +from become_yukarin import VoiceChanger from become_yukarin.config.config import create_from_json as create_conv_config from become_yukarin.config.sr_config import create_from_json as create_sr_config from become_yukarin.dataset.dataset import AcousticFeatureProcess @@ -68,9 +69,10 @@ def process(p: Path, acoustic_converter: AcousticConverter, super_resolution: Su input_wave = acoustic_converter._wave_process(str(p), test=True) input_feature = acoustic_converter._feature_process(input_wave, test=True) converted_feature = acoustic_converter.convert_to_feature(input=input_feature, out_sampling_rate=param.voice_param.sample_rate) - # converted_feature.f0 = input_feature.f0 + # # converted_feature.f0 = input_feature.f0 + + wave = super_resolution(converted_feature.spectrogram.astype(numpy.float64), acoustic_feature=converted_feature, sampling_rate=param.voice_param.sample_rate) - wave = super_resolution(converted_feature.spectrogram, acoustic_feature=converted_feature, sampling_rate=param.voice_param.sample_rate) librosa.output.write_wav(str(output / p.stem) + '.wav', wave.wave, wave.sampling_rate, norm=True) except: import traceback @@ -82,6 +84,7 @@ def run(): acoustic_converter = load_acoustic_converter() super_resolution = load_super_resolution() + # voice_changer = VoiceChanger(acoustic_converter: acoustic_converter, super_resolution: super_resolution) output = output_directory.absolute() output.mkdir(exist_ok=True) |