diff --git a/speechx/examples/feat/linear_spectrogram_main.cc b/speechx/examples/feat/linear_spectrogram_main.cc
index e1f0a89548ad0362950a758ceed445f2c61ee9d8..c29d2b21fd0e58192bc62e1f55110f691eb56d91 100644
--- a/speechx/examples/feat/linear_spectrogram_main.cc
+++ b/speechx/examples/feat/linear_spectrogram_main.cc
@@ -20,7 +20,7 @@
 #include "frontend/audio_cache.h"
 #include "frontend/data_cache.h"
 #include "frontend/feature_cache.h"
-#include "frontend/feature_extractor_interface.h"
+#include "frontend/frontend_itf.h"
 #include "frontend/normalizer.h"
 #include "kaldi/feat/wave-reader.h"
 #include "kaldi/util/kaldi-io.h"
@@ -170,13 +170,13 @@ int main(int argc, char* argv[]) {
     // feature pipeline: wave cache --> decibel_normalizer --> hanning
     // window -->linear_spectrogram --> global cmvn -> feat cache
 
-    // std::unique_ptr<ppspeech::FeatureExtractorInterface> data_source(new
+    // std::unique_ptr<ppspeech::FrontendInterface> data_source(new
     // ppspeech::DataCache());
-    std::unique_ptr<ppspeech::FeatureExtractorInterface> data_source(
+    std::unique_ptr<ppspeech::FrontendInterface> data_source(
         new ppspeech::AudioCache());
 
     ppspeech::DecibelNormalizerOptions db_norm_opt;
-    std::unique_ptr<ppspeech::FeatureExtractorInterface> db_norm(
+    std::unique_ptr<ppspeech::FrontendInterface> db_norm(
         new ppspeech::DecibelNormalizer(db_norm_opt, std::move(data_source)));
 
     ppspeech::LinearSpectrogramOptions opt;
@@ -185,10 +185,10 @@ int main(int argc, char* argv[]) {
     LOG(INFO) << "frame length (ms): " << opt.frame_opts.frame_length_ms;
     LOG(INFO) << "frame shift (ms): " << opt.frame_opts.frame_shift_ms;
 
-    std::unique_ptr<ppspeech::FeatureExtractorInterface> linear_spectrogram(
+    std::unique_ptr<ppspeech::FrontendInterface> linear_spectrogram(
         new ppspeech::LinearSpectrogram(opt, std::move(db_norm)));
 
-    std::unique_ptr<ppspeech::FeatureExtractorInterface> cmvn(
+    std::unique_ptr<ppspeech::FrontendInterface> cmvn(
         new ppspeech::CMVN(FLAGS_cmvn_write_path,
                            std::move(linear_spectrogram)));
 
diff --git a/speechx/speechx/frontend/audio_cache.h b/speechx/speechx/frontend/audio_cache.h
index b6c82c69e13bac6cce47bb421724c50b2b43267b..f48da12b7d77052980cbe42afd6cefdf090f8288 100644
--- a/speechx/speechx/frontend/audio_cache.h
+++ b/speechx/speechx/frontend/audio_cache.h
@@ -16,12 +16,12 @@
 #pragma once
 
 #include "base/common.h"
-#include "frontend/feature_extractor_interface.h"
+#include "frontend/frontend_itf.h"
 
 namespace ppspeech {
 
 // waves cache
-class AudioCache : public FeatureExtractorInterface {
+class AudioCache : public FrontendInterface {
   public:
     explicit AudioCache(int buffer_size = kint16max);
 
diff --git a/speechx/speechx/frontend/data_cache.h b/speechx/speechx/frontend/data_cache.h
index dea51d76ee63cde1cae9fc87e3c04c321e5538e0..b8ce6bf65ee311c4f414ae27b0045d437ebcec38 100644
--- a/speechx/speechx/frontend/data_cache.h
+++ b/speechx/speechx/frontend/data_cache.h
@@ -17,13 +17,13 @@
 
 
 #include "base/common.h"
-#include "frontend/feature_extractor_interface.h"
+#include "frontend/frontend_itf.h"
 
 
 namespace ppspeech {
 // A data source for testing different frontend module.
 // It accepts waves or feats.
-class DataCache : public FeatureExtractorInterface {
+class DataCache : public FrontendInterface {
   public:
     explicit DataCache() { finished_ = false; }
 
diff --git a/speechx/speechx/frontend/fbank.h b/speechx/speechx/frontend/fbank.h
index 7d9cf4221254acb1f0eec4430b4f778cd8e1f255..68267b3d0efee480c89b0816489c6014baceb13f 100644
--- a/speechx/speechx/frontend/fbank.h
+++ b/speechx/speechx/frontend/fbank.h
@@ -20,10 +20,10 @@
 
 namespace ppspeech {
 
-class FbankExtractor : FeatureExtractorInterface {
+class FbankExtractor : FrontendInterface {
   public:
     explicit FbankExtractor(const FbankOptions& opts,
-                            share_ptr<FeatureExtractorInterface> pre_extractor);
+                            share_ptr<FrontendInterface> pre_extractor);
     virtual void AcceptWaveform(
         const kaldi::Vector<kaldi::BaseFloat>& input) = 0;
     virtual void Read(kaldi::Vector<kaldi::BaseFloat>* feat) = 0;
diff --git a/speechx/speechx/frontend/feature_cache.cc b/speechx/speechx/frontend/feature_cache.cc
index dad6907ce4985563682d80039d500450e8c96ccc..53b7076d56d7db08c5f1ade5d2f896bf87c6c0d2 100644
--- a/speechx/speechx/frontend/feature_cache.cc
+++ b/speechx/speechx/frontend/feature_cache.cc
@@ -24,7 +24,7 @@ using kaldi::SubVector;
 using std::unique_ptr;
 
 FeatureCache::FeatureCache(
-    int max_size, unique_ptr<FeatureExtractorInterface> base_extractor) {
+    int max_size, unique_ptr<FrontendInterface> base_extractor) {
     max_size_ = max_size;
     base_extractor_ = std::move(base_extractor);
 }
diff --git a/speechx/speechx/frontend/feature_cache.h b/speechx/speechx/frontend/feature_cache.h
index f52b9b0f6943096c399a7ce8f79076fa5efa72d4..1281ec35a4652f99d26f388b9f786cccb9107e32 100644
--- a/speechx/speechx/frontend/feature_cache.h
+++ b/speechx/speechx/frontend/feature_cache.h
@@ -15,15 +15,15 @@
 #pragma once
 
 #include "base/common.h"
-#include "frontend/feature_extractor_interface.h"
+#include "frontend/frontend_itf.h"
 
 namespace ppspeech {
 
-class FeatureCache : public FeatureExtractorInterface {
+class FeatureCache : public FrontendInterface {
   public:
     explicit FeatureCache(
         int32 max_size = kint16max,
-        std::unique_ptr<FeatureExtractorInterface> base_extractor = NULL);
+        std::unique_ptr<FrontendInterface> base_extractor = NULL);
 
     // Feed feats or waves
     virtual void Accept(const kaldi::VectorBase<kaldi::BaseFloat>& inputs);
@@ -53,7 +53,7 @@ class FeatureCache : public FeatureExtractorInterface {
     bool Compute();
 
     size_t max_size_;
-    std::unique_ptr<FeatureExtractorInterface> base_extractor_;
+    std::unique_ptr<FrontendInterface> base_extractor_;
 
     std::mutex mutex_;
     std::queue<kaldi::Vector<BaseFloat>> cache_;
diff --git a/speechx/speechx/frontend/feature_extractor_controller.h b/speechx/speechx/frontend/feature_extractor_controller.h
deleted file mode 100644
index 0544a1e298b8e7dc871d13f546398a5c28308b0e..0000000000000000000000000000000000000000
--- a/speechx/speechx/frontend/feature_extractor_controller.h
+++ /dev/null
@@ -1,13 +0,0 @@
-// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
diff --git a/speechx/speechx/frontend/feature_extractor_controller_impl.h b/speechx/speechx/frontend/feature_extractor_controller_impl.h
deleted file mode 100644
index 0544a1e298b8e7dc871d13f546398a5c28308b0e..0000000000000000000000000000000000000000
--- a/speechx/speechx/frontend/feature_extractor_controller_impl.h
+++ /dev/null
@@ -1,13 +0,0 @@
-// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
diff --git a/speechx/speechx/frontend/feature_extractor_interface.h b/speechx/speechx/frontend/frontend_itf.h
similarity index 97%
rename from speechx/speechx/frontend/feature_extractor_interface.h
rename to speechx/speechx/frontend/frontend_itf.h
index 5da2526b98c2107992936b0736837feb2ec47652..7913cc7c086564bd01981f46eac5701a1a799ff1 100644
--- a/speechx/speechx/frontend/feature_extractor_interface.h
+++ b/speechx/speechx/frontend/frontend_itf.h
@@ -19,7 +19,7 @@
 
 namespace ppspeech {
 
-class FeatureExtractorInterface {
+class FrontendInterface {
   public:
     // Feed inputs: features(2D saved in 1D) or waveforms(1D).
     virtual void Accept(const kaldi::VectorBase<kaldi::BaseFloat>& inputs) = 0;
diff --git a/speechx/speechx/frontend/linear_spectrogram.cc b/speechx/speechx/frontend/linear_spectrogram.cc
index 41bc8743a939e4f45b5607d71435e246fc480ca7..2ba00785aa70f578025f2b59ec233c4acb855e83 100644
--- a/speechx/speechx/frontend/linear_spectrogram.cc
+++ b/speechx/speechx/frontend/linear_spectrogram.cc
@@ -27,7 +27,7 @@ using std::vector;
 
 LinearSpectrogram::LinearSpectrogram(
     const LinearSpectrogramOptions& opts,
-    std::unique_ptr<FeatureExtractorInterface> base_extractor) {
+    std::unique_ptr<FrontendInterface> base_extractor) {
     opts_ = opts;
     base_extractor_ = std::move(base_extractor);
     int32 window_size = opts.frame_opts.WindowSize();
diff --git a/speechx/speechx/frontend/linear_spectrogram.h b/speechx/speechx/frontend/linear_spectrogram.h
index 10853904d1dc3417938973aa7ea5018c9e5ae590..136441efeec99d6936d73043e4147c5783113779 100644
--- a/speechx/speechx/frontend/linear_spectrogram.h
+++ b/speechx/speechx/frontend/linear_spectrogram.h
@@ -16,7 +16,7 @@
 #pragma once
 
 #include "base/common.h"
-#include "frontend/feature_extractor_interface.h"
+#include "frontend/frontend_itf.h"
 #include "kaldi/feat/feature-window.h"
 
 namespace ppspeech {
@@ -35,11 +35,11 @@ struct LinearSpectrogramOptions {
     }
 };
 
-class LinearSpectrogram : public FeatureExtractorInterface {
+class LinearSpectrogram : public FrontendInterface {
   public:
     explicit LinearSpectrogram(
         const LinearSpectrogramOptions& opts,
-        std::unique_ptr<FeatureExtractorInterface> base_extractor);
+        std::unique_ptr<FrontendInterface> base_extractor);
     virtual void Accept(const kaldi::VectorBase<kaldi::BaseFloat>& inputs);
     virtual bool Read(kaldi::Vector<kaldi::BaseFloat>* feats);
     // the dim_ is the dim of single frame feature
@@ -61,7 +61,7 @@ class LinearSpectrogram : public FeatureExtractorInterface {
     std::vector<kaldi::BaseFloat> hanning_window_;
     kaldi::BaseFloat hanning_window_energy_;
     LinearSpectrogramOptions opts_;
-    std::unique_ptr<FeatureExtractorInterface> base_extractor_;
+    std::unique_ptr<FrontendInterface> base_extractor_;
     int chunk_sample_size_;
     DISALLOW_COPY_AND_ASSIGN(LinearSpectrogram);
 };
diff --git a/speechx/speechx/frontend/normalizer.cc b/speechx/speechx/frontend/normalizer.cc
index 524125619a74f8e12b7d9932b7529a4ab3bfde0f..26f11b692d2d5c022e26860f2b6f693f904b0741 100644
--- a/speechx/speechx/frontend/normalizer.cc
+++ b/speechx/speechx/frontend/normalizer.cc
@@ -28,7 +28,7 @@ using std::unique_ptr;
 
 DecibelNormalizer::DecibelNormalizer(
     const DecibelNormalizerOptions& opts,
-    std::unique_ptr<FeatureExtractorInterface> base_extractor) {
+    std::unique_ptr<FrontendInterface> base_extractor) {
     base_extractor_ = std::move(base_extractor);
     opts_ = opts;
     dim_ = 1;
@@ -92,7 +92,7 @@ bool DecibelNormalizer::Compute(VectorBase<BaseFloat>* waves) const {
 }
 
 CMVN::CMVN(std::string cmvn_file,
-           unique_ptr<FeatureExtractorInterface> base_extractor)
+           unique_ptr<FrontendInterface> base_extractor)
     : var_norm_(true) {
     base_extractor_ = std::move(base_extractor);
     bool binary;
diff --git a/speechx/speechx/frontend/normalizer.h b/speechx/speechx/frontend/normalizer.h
index 352d1e1677e7e9566f3c3ea08d65235ae897a73a..df181961248b4581337eccf4962874dde3e619e7 100644
--- a/speechx/speechx/frontend/normalizer.h
+++ b/speechx/speechx/frontend/normalizer.h
@@ -16,7 +16,7 @@
 #pragma once
 
 #include "base/common.h"
-#include "frontend/feature_extractor_interface.h"
+#include "frontend/frontend_itf.h"
 #include "kaldi/matrix/kaldi-matrix.h"
 #include "kaldi/util/options-itf.h"
 
@@ -40,11 +40,11 @@ struct DecibelNormalizerOptions {
     }
 };
 
-class DecibelNormalizer : public FeatureExtractorInterface {
+class DecibelNormalizer : public FrontendInterface {
   public:
     explicit DecibelNormalizer(
         const DecibelNormalizerOptions& opts,
-        std::unique_ptr<FeatureExtractorInterface> base_extractor);
+        std::unique_ptr<FrontendInterface> base_extractor);
     virtual void Accept(const kaldi::VectorBase<kaldi::BaseFloat>& waves);
     virtual bool Read(kaldi::Vector<kaldi::BaseFloat>* waves);
     // noramlize audio, the dim is 1.
@@ -57,15 +57,15 @@ class DecibelNormalizer : public FeatureExtractorInterface {
     bool Compute(kaldi::VectorBase<kaldi::BaseFloat>* waves) const;
     DecibelNormalizerOptions opts_;
     size_t dim_;
-    std::unique_ptr<FeatureExtractorInterface> base_extractor_;
+    std::unique_ptr<FrontendInterface> base_extractor_;
     kaldi::Vector<kaldi::BaseFloat> waveform_;
 };
 
 
-class CMVN : public FeatureExtractorInterface {
+class CMVN : public FrontendInterface {
   public:
     explicit CMVN(std::string cmvn_file,
-                  std::unique_ptr<FeatureExtractorInterface> base_extractor);
+                  std::unique_ptr<FrontendInterface> base_extractor);
     virtual void Accept(const kaldi::VectorBase<kaldi::BaseFloat>& inputs);
 
     // the length of feats = feature_row * feature_dim,
@@ -81,7 +81,7 @@ class CMVN : public FeatureExtractorInterface {
     void Compute(kaldi::VectorBase<kaldi::BaseFloat>* feats) const;
     void ApplyCMVN(kaldi::MatrixBase<BaseFloat>* feats);
     kaldi::Matrix<double> stats_;
-    std::unique_ptr<FeatureExtractorInterface> base_extractor_;
+    std::unique_ptr<FrontendInterface> base_extractor_;
     size_t dim_;
     bool var_norm_;
 };
diff --git a/speechx/speechx/nnet/decodable.cc b/speechx/speechx/nnet/decodable.cc
index e6315d07a670e5a555ada56aa5014d394b4b98c7..542168d245950c253f42ed9d87014646d547eb67 100644
--- a/speechx/speechx/nnet/decodable.cc
+++ b/speechx/speechx/nnet/decodable.cc
@@ -22,7 +22,7 @@ using std::vector;
 using kaldi::Vector;
 
 Decodable::Decodable(const std::shared_ptr<NnetInterface>& nnet,
-                     const std::shared_ptr<FeatureExtractorInterface>& frontend)
+                     const std::shared_ptr<FrontendInterface>& frontend)
     : frontend_(frontend), nnet_(nnet), frame_offset_(0), frames_ready_(0) {}
 
 void Decodable::Acceptlikelihood(const Matrix<BaseFloat>& likelihood) {
diff --git a/speechx/speechx/nnet/decodable.h b/speechx/speechx/nnet/decodable.h
index 7938b5823e8aa4e75816c025a79e94816a96ba04..ef17601fac992554fae25428d9a2d27063c3da10 100644
--- a/speechx/speechx/nnet/decodable.h
+++ b/speechx/speechx/nnet/decodable.h
@@ -13,7 +13,7 @@
 // limitations under the License.
 
 #include "base/common.h"
-#include "frontend/feature_extractor_interface.h"
+#include "frontend/frontend_itf.h"
 #include "kaldi/matrix/kaldi-matrix.h"
 #include "nnet/decodable-itf.h"
 #include "nnet/nnet_interface.h"
@@ -26,7 +26,7 @@ class Decodable : public kaldi::DecodableInterface {
   public:
     explicit Decodable(
         const std::shared_ptr<NnetInterface>& nnet,
-        const std::shared_ptr<FeatureExtractorInterface>& frontend);
+        const std::shared_ptr<FrontendInterface>& frontend);
     // void Init(DecodableOpts config);
     virtual kaldi::BaseFloat LogLikelihood(int32 frame, int32 index);
     virtual bool IsLastFrame(int32 frame) const;
@@ -41,7 +41,7 @@ class Decodable : public kaldi::DecodableInterface {
 
   private:
     bool AdvanceChunk();
-    std::shared_ptr<FeatureExtractorInterface> frontend_;
+    std::shared_ptr<FrontendInterface> frontend_;
     std::shared_ptr<NnetInterface> nnet_;
     kaldi::Matrix<kaldi::BaseFloat> nnet_cache_;
     // std::vector<std::vector<kaldi::BaseFloat>> nnet_cache_;