From b27d67d2fbe2684ba8f19728ae089f038822192d Mon Sep 17 00:00:00 2001
From: Fangjun Kuang <csukuangfj@gmail.com>
Date: Tue, 28 Sep 2021 08:18:06 +0800
Subject: [PATCH] Preserve cut id in mix cut transformer.

---
 egs/librispeech/ASR/tdnn_lstm_ctc/asr_datamodule.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/egs/librispeech/ASR/tdnn_lstm_ctc/asr_datamodule.py b/egs/librispeech/ASR/tdnn_lstm_ctc/asr_datamodule.py
index 8290e71d1..d3eab87a9 100644
--- a/egs/librispeech/ASR/tdnn_lstm_ctc/asr_datamodule.py
+++ b/egs/librispeech/ASR/tdnn_lstm_ctc/asr_datamodule.py
@@ -162,7 +162,9 @@ class LibriSpeechAsrDataModule(DataModule):
         cuts_musan = load_manifest(self.args.feature_dir / "cuts_musan.json.gz")
 
         logging.info("About to create train dataset")
-        transforms = [CutMix(cuts=cuts_musan, prob=0.5, snr=(10, 20))]
+        transforms = [
+            CutMix(cuts=cuts_musan, prob=0.5, snr=(10, 20), preserve_id=True)
+        ]
         if self.args.concatenate_cuts:
             logging.info(
                 f"Using cut concatenation with duration factor "