eschmidbauer/DL-Art-School
1
0
Fork 0
forked from mrq/DL-Art-School
Code Issues Pull Requests Packages Projects Releases Wiki Activity

f

Browse Source
This commit is contained in:
James Betker 2022-02-11 11:05:13 -07:00
parent 40b08a52d0
commit 102142d1eb
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
codes/trainer/eval/audio_diffusion_fid.py
View File

@ -15,7 +15,7 @@ import trainer.eval.evaluator as evaluator
from data.audio.paired_voice_audio_dataset import load_tsv_aligned_codes from data.audio.paired_voice_audio_dataset import load_tsv_aligned_codes
from data.audio.unsupervised_audio_dataset import load_audio from data.audio.unsupervised_audio_dataset import load_audio
from scripts.audio.gen.speech_synthesis_utils import load_discrete_vocoder_diffuser from scripts.audio.gen.speech_synthesis_utils import load_discrete_vocoder_diffuser
from utils.util import opt_get from utils.util import ceil_multiple, opt_get
class AudioDiffusionFid(evaluator.Evaluator): class AudioDiffusionFid(evaluator.Evaluator):
@ -109,11 +109,11 @@ class AudioDiffusionFid(evaluator.Evaluator):
if __name__ == '__main__': if __name__ == '__main__':
from utils.util import load_model_from_config, ceil_multiple, opt_get from utils.util import load_model_from_config
diffusion = load_model_from_config('X:\\dlas\\experiments\\train_diffusion_tts5_medium.yml', 'generator', diffusion = load_model_from_config('X:\\dlas\\experiments\\train_diffusion_tts5_medium.yml', 'generator',
also_load_savepoint=False, load_path='X:\\dlas\\experiments\\train_diffusion_tts5_medium\\models\\73000_generator_ema.pth').cuda() also_load_savepoint=False, load_path='X:\\dlas\\experiments\\train_diffusion_tts5_medium\\models\\73000_generator_ema.pth').cuda()
opt_eval = {'eval_tsv': 'Y:\\libritts\\test-clean\\transcribed-brief-w2v.tsv', 'diffusion_steps': 50} opt_eval = {'eval_tsv': 'Y:\\libritts\\test-clean\\transcribed-brief-w2v.tsv', 'diffusion_steps': 50}
env = {'rank': 0, 'base_path': 'D:\\tmp\\test_eval', 'step': 500, 'device': 'cuda'} env = {'rank': 0, 'base_path': 'D:\\tmp\\test_eval', 'step': 500, 'device': 'cuda'}
eval = StyleTransferEvaluator(diffusion, opt_eval, env) eval = AudioDiffusionFid(diffusion, opt_eval, env)
eval.perform_eval() eval.perform_eval()