diff --git a/egs/dihard_2018/v1/run.sh b/egs/dihard_2018/v1/run.sh
index 53b4ee569f4..429a1231975 100755
--- a/egs/dihard_2018/v1/run.sh
+++ b/egs/dihard_2018/v1/run.sh
@@ -44,7 +44,7 @@ if [ $stage -le 1 ]; then
   # Make MFCCs for each dataset
   for name in train dihard_2018_dev dihard_2018_eval; do
     steps/make_mfcc.sh --write-utt2num-frames true \
-      --mfcc-config conf/mfcc.conf --nj 40 --cmd "$train_cmd" \
+      --mfcc-config conf/mfcc.conf --nj 40 --cmd "$train_cmd --max-jobs-run 20" \
       data/${name} exp/make_mfcc $mfccdir
     utils/fix_data_dir.sh data/${name}
   done
diff --git a/egs/dihard_2018/v2/run.sh b/egs/dihard_2018/v2/run.sh
index d39496e4f7d..ab1dca22cfd 100755
--- a/egs/dihard_2018/v2/run.sh
+++ b/egs/dihard_2018/v2/run.sh
@@ -42,7 +42,7 @@ fi
 if [ $stage -le 1 ]; then
   # Make MFCCs for each dataset.
   for name in train dihard_2018_dev dihard_2018_eval; do
-    steps/make_mfcc.sh --write-utt2num-frames true --mfcc-config conf/mfcc.conf --nj 40 --cmd "$train_cmd" \
+    steps/make_mfcc.sh --write-utt2num-frames true --mfcc-config conf/mfcc.conf --nj 40 --cmd "$train_cmd --max-jobs-run 20" \
       data/${name} exp/make_mfcc $mfccdir
     utils/fix_data_dir.sh data/${name}
   done
@@ -141,7 +141,7 @@ if [ $stage -le 3 ]; then
   # Make MFCCs for the augmented data.  Note that we do not compute a new
   # vad.scp file here.  Instead, we use the vad.scp from the clean version of
   # the list.
-  steps/make_mfcc.sh --mfcc-config conf/mfcc.conf --nj 40 --cmd "$train_cmd" \
+  steps/make_mfcc.sh --mfcc-config conf/mfcc.conf --nj 40 --cmd "$train_cmd --max-jobs-run 20" \
     data/train_aug_1m exp/make_mfcc $mfccdir
 
   # Combine the clean and augmented training data.  This is now roughly