speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge pull request #73 from alibaba-damo-academy/main

Browse Source 
fix bug, batch cif predictor tail
This commit is contained in:
zhifu gao 2023-02-08 13:21:12 +08:00 committed by GitHub
commit 8871dcb93a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 3 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
funasr/models/predictor/cif.py
View File

@ -208,7 +208,8 @@ class CifPredictorV2(nn.Module):
mask_2 = torch.cat([ones_t, mask], dim=1) mask_2 = torch.cat([ones_t, mask], dim=1)
mask = mask_2 - mask_1 mask = mask_2 - mask_1
tail_threshold = mask * tail_threshold tail_threshold = mask * tail_threshold
alphas = torch.cat([alphas, tail_threshold], dim=1) alphas = torch.cat([alphas, zeros_t], dim=1)
alphas = torch.add(alphas, tail_threshold)
else: else:
tail_threshold = torch.tensor([tail_threshold], dtype=alphas.dtype).to(alphas.device) tail_threshold = torch.tensor([tail_threshold], dtype=alphas.dtype).to(alphas.device)
tail_threshold = torch.reshape(tail_threshold, (1, 1)) tail_threshold = torch.reshape(tail_threshold, (1, 1))
@ -654,4 +655,4 @@ class CifPredictorV3(nn.Module):
predictor_alignments = index_div_bool_zeros_count_tile_out predictor_alignments = index_div_bool_zeros_count_tile_out
predictor_alignments_length = predictor_alignments.sum(-1).type(encoder_sequence_length.dtype) predictor_alignments_length = predictor_alignments.sum(-1).type(encoder_sequence_length.dtype)
return predictor_alignments.detach(), predictor_alignments_length.detach() return predictor_alignments.detach(), predictor_alignments_length.detach()