Skip to content

Commit

Permalink
fix current length
Browse files Browse the repository at this point in the history
  • Loading branch information
Mddct committed Aug 14, 2024
1 parent e33da5f commit 5288f24
Showing 1 changed file with 6 additions and 1 deletion.
7 changes: 6 additions & 1 deletion wenet/ssl/init_dataset.py
Original file line number Diff line number Diff line change
Expand Up @@ -33,16 +33,21 @@ def __init__(

def __iter__(self):
for elem in self.dp:
current_length = self.length + self.length_fn(elem)
elem_length = self.length_fn(elem)
current_length = self.length + elem_length
if current_length >= self.max_length:
long_elem = self.merge_fn(self.buf)
yield long_elem
del self.buf
self.buf = []
self.length = 0
self.buf.append(elem)
self.length += current_length
if len(self.buf) > 0:
yield self.merge_fn(self.buf)
del self.buf
self.buf = []
self.length = 0


def cat_speech(buffer: List):
Expand Down

0 comments on commit 5288f24

Please sign in to comment.