From 8e0b7ea518a62b2b89789161af4a57bac6cda3f7 Mon Sep 17 00:00:00 2001
From: Mingshuang Luo <37799481+luomingshuang@users.noreply.github.com>
Date: Mon, 4 Jul 2022 11:59:37 +0800
Subject: [PATCH] mv split cuts before computing feature (#461)

---
 .../ASR/local/compute_fbank_wenetspeech_dev_test.py  |  9 +++++----
 .../ASR/local/compute_fbank_wenetspeech_splits.py    | 12 +++++-------
 2 files changed, 10 insertions(+), 11 deletions(-)

diff --git a/egs/wenetspeech/ASR/local/compute_fbank_wenetspeech_dev_test.py b/egs/wenetspeech/ASR/local/compute_fbank_wenetspeech_dev_test.py
index 7efcbd4fd..c10a84d21 100755
--- a/egs/wenetspeech/ASR/local/compute_fbank_wenetspeech_dev_test.py
+++ b/egs/wenetspeech/ASR/local/compute_fbank_wenetspeech_dev_test.py
@@ -63,8 +63,12 @@ def compute_fbank_wenetspeech_dev_test():
         logging.info(f"Loading {raw_cuts_path}")
         cut_set = CutSet.from_file(raw_cuts_path)
 
-        logging.info("Computing features")
+        logging.info("Splitting cuts into smaller chunks")
+        cut_set = cut_set.trim_to_supervisions(
+            keep_overlapping=False, min_duration=None
+        )
 
+        logging.info("Computing features")
         cut_set = cut_set.compute_and_store_features_batch(
             extractor=extractor,
             storage_path=f"{in_out_dir}/feats_{partition}",
@@ -72,9 +76,6 @@ def compute_fbank_wenetspeech_dev_test():
             batch_duration=batch_duration,
             storage_type=LilcomHdf5Writer,
         )
-        cut_set = cut_set.trim_to_supervisions(
-            keep_overlapping=False, min_duration=None
-        )
 
         logging.info(f"Saving to {cuts_path}")
         cut_set.to_file(cuts_path)
diff --git a/egs/wenetspeech/ASR/local/compute_fbank_wenetspeech_splits.py b/egs/wenetspeech/ASR/local/compute_fbank_wenetspeech_splits.py
index 4622bdb55..bf9a03b20 100755
--- a/egs/wenetspeech/ASR/local/compute_fbank_wenetspeech_splits.py
+++ b/egs/wenetspeech/ASR/local/compute_fbank_wenetspeech_splits.py
@@ -128,8 +128,12 @@ def compute_fbank_wenetspeech_splits(args):
         logging.info(f"Loading {raw_cuts_path}")
         cut_set = CutSet.from_file(raw_cuts_path)
 
-        logging.info("Computing features")
+        logging.info("Splitting cuts into smaller chunks.")
+        cut_set = cut_set.trim_to_supervisions(
+            keep_overlapping=False, min_duration=None
+        )
 
+        logging.info("Computing features")
         cut_set = cut_set.compute_and_store_features_batch(
             extractor=extractor,
             storage_path=f"{output_dir}/feats_{subset}_{idx}",
@@ -138,14 +142,8 @@ def compute_fbank_wenetspeech_splits(args):
             storage_type=LilcomChunkyWriter,
         )
 
-        logging.info("About to split cuts into smaller chunks.")
-        cut_set = cut_set.trim_to_supervisions(
-            keep_overlapping=False, min_duration=None
-        )
-
         logging.info(f"Saving to {cuts_path}")
         cut_set.to_file(cuts_path)
-        logging.info(f"Saved to {cuts_path}")
 
 
 def main():