diff --git a/README.md b/README.md
index 7ceae17f2..eb16dd587 100644
--- a/README.md
+++ b/README.md
@@ -506,7 +506,7 @@ VoxForge European       |   31.21           |   20.47
 VoxForge Indian         |   56.79           |   28.15
 Baidu Internal Testset  |   47.73           |   8.92
 
-For reproducing benchmark results on VoxForge data, we provide a script to generate VoxForge dialect manifest files. Please go to ```data/voxforge``` and execute ```sh run_data.sh``` to get VoxForge dialect manifest files. Notice that VoxForge data may keep updated and the generated manifest files may have difference from those we evaluated.
+For reproducing benchmark results on VoxForge data, we provide a script to download data and generate VoxForge dialect manifest files. Please go to ```data/voxforge``` and execute ```sh run_data.sh``` to get VoxForge dialect manifest files. Notice that VoxForge data may keep updating and the generated manifest files may have difference from those we evaluated on.
 
 #### Benchmark Results for Mandarin Model (Character Error Rate)
 
diff --git a/data/voxforge/run_data.sh b/data/voxforge/run_data.sh
index e0a9f1b3f..c6ff71118 100644
--- a/data/voxforge/run_data.sh
+++ b/data/voxforge/run_data.sh
@@ -1,10 +1,8 @@
 #! /usr/bin/env bash
 
-cd ../.. > /dev/null
-
 # download data, generate manifests
-PYTHONPATH=.:$PYTHONPATH python data/voxforge/voxforge.py \
---manifest_prefix='data/voxforge/manifest' \
+PYTHONPATH=../../:$PYTHONPATH python voxforge.py \
+--manifest_prefix='./manifest' \
 --target_dir='~/.cache/paddle/dataset/speech/VoxForge' \
 --is_merge_dialect=True \
 --dialects 'american' 'british' 'australian' 'european' 'irish' 'canadian' 'indian'