From 678ad2b8a9dd152c935cda399135baeb38511135 Mon Sep 17 00:00:00 2001
From: jinzr <zengrui.jin0@gmail.com>
Date: Fri, 15 Mar 2024 10:49:12 +0800
Subject: [PATCH] Update preprocess_commonvoice.py

---
 egs/commonvoice/ASR/local/preprocess_commonvoice.py | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/egs/commonvoice/ASR/local/preprocess_commonvoice.py b/egs/commonvoice/ASR/local/preprocess_commonvoice.py
index d41af4015..3be85ed15 100755
--- a/egs/commonvoice/ASR/local/preprocess_commonvoice.py
+++ b/egs/commonvoice/ASR/local/preprocess_commonvoice.py
@@ -93,6 +93,10 @@ def normalize_text(utt: str, language: str) -> str:
             .replace("～", "")
             .replace("；", "")
             .replace("", "")
+            .replace("﹔", "")
+            .replace("／", "")
+            .replace("Ａ", "")
+            .replace("Ｂ", "")
             .upper()
         )
     else: