diff --git a/examples/voxceleb/v2/README.md b/examples/voxceleb/v2/README.md
index 679c6920..45396a0e 100644
--- a/examples/voxceleb/v2/README.md
+++ b/examples/voxceleb/v2/README.md
@@ -50,7 +50,6 @@ If you are interested in the PLDA scoring (which is inferior to the simple cosin
 ```bash
 local/score_plda.sh --stage 1 --stop-stage 3 --exp_dir exp_name
 ```
-Note that you need to prepare an additional utt2utt file in the vox1 data directory since each speaker is enrolled with only one utterance.
 
 The results on ResNet293 (large margin, no asnorm) are:
 
diff --git a/examples/voxceleb/v2/local/score_plda.sh b/examples/voxceleb/v2/local/score_plda.sh
old mode 100644
new mode 100755
index 46ec4d42..7354094c
--- a/examples/voxceleb/v2/local/score_plda.sh
+++ b/examples/voxceleb/v2/local/score_plda.sh
@@ -47,7 +47,7 @@ if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
       --exp_dir ${exp_dir} \
       --enroll_scp_path ${exp_dir}/embeddings/vox1/xvector.scp \
       --test_scp_path ${exp_dir}/embeddings/vox1/xvector.scp \
-      --utt2spk ${data}/vox1/utt2utt \
+      --utt2spk <(cat ${data}/vox1/utt2spk | awk '{print $1, $1}') \
       --trial ${trials_dir}/${x}
   done
 fi
@@ -67,4 +67,4 @@ if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
     python wespeaker/bin/compute_det.py \
         ${scores_dir}/${x}.pldascore
   done
-fi
\ No newline at end of file
+fi
diff --git a/wespeaker/bin/compute_det.py b/wespeaker/bin/compute_det.py
index 07dd1283..59861750 100644
--- a/wespeaker/bin/compute_det.py
+++ b/wespeaker/bin/compute_det.py
@@ -40,7 +40,7 @@ def compute_det(scores_file, det_file):
 
 def main(*scores_files):
     for scores_file in scores_files:
-        det_file = scores_file[:-6] + ".det.png"
+        det_file = scores_file + ".det.png"
         compute_det(scores_file, det_file)