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

modify unit test for speech_diarization_sond-en-us-callhome-8k-n16k4-pytorch

Browse Source
This commit is contained in:
志浩 2023-03-09 17:24:41 +08:00
parent 77cbfde968
commit cf41c0aa3a
1 changed files with 5 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
funasr/models/pooling/statistic_pooling.py
View File

@ -82,13 +82,16 @@ def windowed_statistic_pooling(
tt = xs_pad.shape[2]
num_chunk = int(math.ceil(tt / pooling_stride))
pad = pooling_size // 2
features = F.pad(xs_pad, (0, 0, pad, pad), "reflect")
if len(xs_pad.shape == 4):
features = F.pad(xs_pad, (0, 0, pad, pad), "reflect")
else:
features = F.pad(xs_pad, (pad, pad), "reflect")
stat_list = []
for i in range(num_chunk):
# B x C
st, ed = i*pooling_stride, i*pooling_stride+pooling_size
stat = statistic_pooling(features[:, :, st: ed, :], pooling_dim=pooling_dim)
stat = statistic_pooling(features[:, :, st: ed], pooling_dim=pooling_dim)
stat_list.append(stat.unsqueeze(2))
# B x C x T