From 46e2b56c6df40f14e7fd97f6eff22cfa68dbd0c2 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E5=BD=AD=E9=9C=87=E4=B8=9C?= <275331498@qq.com> Date: Sun, 19 Jan 2025 17:53:18 +0800 Subject: [PATCH] pad cuts by num_samples as default fix duration precision issue: https://github.com/lhotse-speech/lhotse/blob/509f4a97ab79a9eaeb8ba377d2a8e4acc794e45b/lhotse/cut/set.py#L2967 0.15052083333333321 * 24000 = 3,612.5 = 3,613 0.15052083 * 24000 = 3,612.49992 = 3,612 --- lhotse/dataset/collation.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/lhotse/dataset/collation.py b/lhotse/dataset/collation.py index 0d7816cad..39a294802 100644 --- a/lhotse/dataset/collation.py +++ b/lhotse/dataset/collation.py @@ -189,7 +189,7 @@ def collate_audio( cuts = maybe_pad( cuts, - duration=max(cut.duration for cut in cuts), + num_samples=max(cut_id2num_samples.values()), direction=pad_direction, preserve_id=True, )