From 00f8371f373805031fee8ab00ab8e1d64a8e1d78 Mon Sep 17 00:00:00 2001
From: Fangjun Kuang <csukuangfj@gmail.com>
Date: Sat, 24 Jul 2021 18:24:04 +0800
Subject: [PATCH] begin to add LM rescoring.

---
 .github/workflows/test.yml     |  2 +-
 egs/librispeech/ASR/prepare.sh | 12 +++++++++++-
 2 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
index f808bffbb..7da954790 100644
--- a/.github/workflows/test.yml
+++ b/.github/workflows/test.yml
@@ -30,7 +30,7 @@ jobs:
     strategy:
       matrix:
         os: [ubuntu-18.04, macos-10.15]
-        python-version: [3.7, 3.9]
+        python-version: [3.6, 3.7, 3.8, 3.9]
         torch: ["1.8.1"]
         k2-version: ["1.2.dev20210723"]
       fail-fast: false
diff --git a/egs/librispeech/ASR/prepare.sh b/egs/librispeech/ASR/prepare.sh
index 3f827b223..87786c5c8 100755
--- a/egs/librispeech/ASR/prepare.sh
+++ b/egs/librispeech/ASR/prepare.sh
@@ -93,13 +93,23 @@ if [ $stage -le 6 ] && [ $stop_stage -ge 6 ]; then
   # We assume you have install kaldilm, if not, please install
   # it using: pip install kaldilm
 
-  if [ ! -e data/lm/G_3_gram.fst.txt ]; then
+  if [ ! -f data/lm/G_3_gram.fst.txt ]; then
+    # It is used in building HLG
     python3 -m kaldilm \
       --read-symbol-table="data/lang/words.txt" \
       --disambig-symbol='#0' \
       --max-order=3 \
       data/lm/3-gram.pruned.1e-7.arpa > data/lm/G_3_gram.fst.txt
   fi
+
+  if [ ! -f data/lm/G_4_gram.fst.txt ]; then
+    # It is used for LM rescoring
+    python3 -m kaldilm \
+      --read-symbol-table="data/lang/words.txt" \
+      --disambig-symbol='#0' \
+      --max-order=4 \
+      data/lm/4-gram.arpa > data/lm/G_4_gram.fst.txt
+  fi
 fi
 
 if [ $stage -le 7 ] && [ $stop_stage -ge 7 ]; then