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

sond pipeline

Browse Source
This commit is contained in:
志浩 2023-02-23 22:05:54 +08:00
parent cb5d258586
commit fc85f34596
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
egs/mars/sd/scripts/real_meeting_process/dump_real_meeting_chunks.py
View File

@ -25,7 +25,7 @@ def calc_rand_ivc(spk, spk2utt, utt2ivc, utt2frames, total_len=3000):
print("Speaker {} has only {} frames, but expect {} frames at least, use them all.".format(spk, count, 300))
short_spk_list.append(spk)
ivc_list = [kaldiio.load_mat(utt2ivc[utt])[np.newaxis, :] for utt in utt_list]
ivc_list = [kaldiio.load_mat(utt2ivc[utt]) for utt in utt_list]
ivc = np.concatenate(ivc_list, axis=0)
ivc = np.mean(ivc, axis=0, keepdims=False)
return ivc
@ -56,7 +56,7 @@ def process(meeting_scp, labels_scp, spk2utt, utt2xvec, utt2frames, meeting2spk_
xvec_list = []
for spk in meeting2spk_list[mid]:
spk_xvec = calc_rand_ivc(spk, spk2utt, utt2xvec, utt2frames, 1000)[np.newaxis, :]
spk_xvec = calc_rand_ivc(spk, spk2utt, utt2xvec, utt2frames, 1000)
xvec_list.append(spk_xvec)
for _ in range(args.n_spk - len(xvec_list)):
xvec_list.append(np.zeros((ivc_dim,), dtype=np.float32))