From 69ab31dbd67f4aafa4c87cad43b98eeaf6efbba2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Piotr=20=C5=BBelasko?= <petezor@gmail.com>
Date: Tue, 23 Jan 2024 09:13:54 -0500
Subject: [PATCH] Fix duplication issues in CutSet.mix() (#1268)

---
 lhotse/cut/set.py           |  1 +
 test/cut/test_cut_mixing.py | 17 +++++++++++++++++
 2 files changed, 18 insertions(+)

diff --git a/lhotse/cut/set.py b/lhotse/cut/set.py
index 09789f67f..267ee16e8 100644
--- a/lhotse/cut/set.py
+++ b/lhotse/cut/set.py
@@ -3391,6 +3391,7 @@ def __iter__(self):
             # or pass it through unchanged.
             if rng.uniform(0.0, 1.0) > self.mix_prob:
                 yield cut
+                continue
             to_mix = next(mix_in_cuts)
             # Determine the SNR - either it's specified or we need to sample one.
             cut_snr = (
diff --git a/test/cut/test_cut_mixing.py b/test/cut/test_cut_mixing.py
index 89ffa4928..f1f31b04d 100644
--- a/test/cut/test_cut_mixing.py
+++ b/test/cut/test_cut_mixing.py
@@ -474,3 +474,20 @@ def test_cut_set_mix_is_lazy():
     mixed = cuts.mix(cuts, snr=10, mix_prob=1.0, seed=0)
 
     assert mixed.is_lazy
+
+
+def test_cut_set_mix_size_is_not_growing():
+    cuts = DummyManifest(CutSet, begin_id=0, end_id=100)
+    noise_cuts = DummyManifest(CutSet, begin_id=10, end_id=20)
+
+    mixed_cuts = cuts.mix(
+        cuts=noise_cuts,
+        duration=None,
+        snr=10,
+        mix_prob=0.1,
+        preserve_id=None,
+        seed=42,
+        random_mix_offset=True,
+    ).to_eager()
+
+    assert len(mixed_cuts) == len(cuts)