deepgeek/LAM_Audio2Expression
2
0
Fork 0
mirror of https://github.com/aigc3d/LAM_Audio2Expression.git synced 2026-02-04 17:39:24 +08:00
Code Issues Packages Projects Releases Wiki Activity

run with cpu

Browse Source
This commit is contained in:
bingochaos
2025-09-26 19:21:53 +08:00
parent aa5bc3487a
commit fe0fe91275
1 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
engines/infer.py
View File

@@ -65,7 +65,7 @@ class InferBase:
n_parameters = sum(p.numel() for p in model.parameters() if p.requires_grad)
self.logger.info(f"Num params: {n_parameters}")
model = create_ddp_model(
model.cuda(),
model,
broadcast_buffers=False,
find_unused_parameters=self.cfg.find_unused_parameters,
)
@@ -117,9 +117,9 @@ class Audio2ExpressionInfer(InferBase):
with torch.no_grad():
input_dict = {}
input_dict['id_idx'] = F.one_hot(torch.tensor(self.cfg.id_idx),
self.cfg.model.backbone.num_identity_classes).cuda(non_blocking=True)[None,...]
self.cfg.model.backbone.num_identity_classes)[None,...]
speech_array, ssr = librosa.load(self.cfg.audio_input, sr=16000)
input_dict['input_audio_array'] = torch.FloatTensor(speech_array).cuda(non_blocking=True)[None,...]
input_dict['input_audio_array'] = torch.FloatTensor(speech_array)[None,...]
end = time.time()
output_dict = self.model(input_dict)
@@ -198,9 +198,9 @@ class Audio2ExpressionInfer(InferBase):
try:
input_dict = {}
input_dict['id_idx'] = F.one_hot(torch.tensor(self.cfg.id_idx),
self.cfg.model.backbone.num_identity_classes).cuda(non_blocking=True)[
self.cfg.model.backbone.num_identity_classes)[
None, ...]
input_dict['input_audio_array'] = torch.FloatTensor(input_audio).cuda(non_blocking=True)[None, ...]
input_dict['input_audio_array'] = torch.FloatTensor(input_audio)[None, ...]
output_dict = self.model(input_dict)
out_exp = output_dict['pred_exp'].squeeze().cpu().numpy()[start_frame:, :]
except: