diff --git a/demos/echo/.gitignore b/demos/echo/.gitignore
new file mode 100644
index 0000000000000000000000000000000000000000..1269488f7fb1f4b56a8c0e5eb48cecbfadfa9219
--- /dev/null
+++ b/demos/echo/.gitignore
@@ -0,0 +1 @@
+data
diff --git a/demos/echo/README.md b/demos/echo/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..10e443f2ed91f96604f4273e07e50f1fdc1ec1df
--- /dev/null
+++ b/demos/echo/README.md
@@ -0,0 +1,3 @@
+# echo system
+
+ASR + TTS
diff --git a/demos/echo/run.sh b/demos/echo/run.sh
new file mode 100755
index 0000000000000000000000000000000000000000..c2379c3a741508460f98a0939f43517f7c0f16ca
--- /dev/null
+++ b/demos/echo/run.sh
@@ -0,0 +1,17 @@
+#!/bin/bash
+
+mkdir -p data
+
+wav_en=data/en.wav
+wav_zh=data/zh.wav
+
+test -e ${wav_en}  || wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/en.wav -P data
+test -e ${wav_zh}  || wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav -P data
+
+pip install paddlehub
+
+asr_en_cmd="import paddlehub as hub; model = hub.Module(name='u2_conformer_librispeech'); print(model.speech_recognize("${wav_en}", device='gpu'))"
+asr_zh_cmd="import paddlehub as hub; model = hub.Module(name='u2_conformer_aishell'); print(model.speech_recognize("${wav_zh}", device='gpu'))"
+
+python -c "${asr_en_cmd}"
+python -c "${asr_zh_cmd}"
\ No newline at end of file