AudioEncoder subclass for G722

BUG=3926
R=henrik.lundin@webrtc.org, kjellander@webrtc.org

Review URL: https://webrtc-codereview.appspot.com/30259004

git-svn-id: http://webrtc.googlecode.com/svn/trunk@7779 4adac7df-926f-26a2-2b94-8c16560cd09d

webrtc/modules/audio_coding/codecs/g722/audio_encoder_g722.cc

@@ -0,0 +1,119 @@
+/*
+ *  Copyright (c) 2014 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "webrtc/modules/audio_coding/codecs/g722/include/audio_encoder_g722.h"
+
+#include <limits>
+#include "webrtc/base/checks.h"
+#include "webrtc/modules/audio_coding/codecs/g722/include/g722_interface.h"
+
+namespace webrtc {
+
+namespace {
+
+const int kSampleRateHz = 16000;
+
+}  // namespace
+
+AudioEncoderG722::EncoderState::EncoderState() {
+  CHECK_EQ(0, WebRtcG722_CreateEncoder(&encoder));
+  CHECK_EQ(0, WebRtcG722_EncoderInit(encoder));
+}
+
+AudioEncoderG722::EncoderState::~EncoderState() {
+  CHECK_EQ(0, WebRtcG722_FreeEncoder(encoder));
+}
+
+AudioEncoderG722::AudioEncoderG722(const Config& config)
+    : num_channels_(config.num_channels),
+      num_10ms_frames_per_packet_(config.frame_size_ms / 10),
+      num_10ms_frames_buffered_(0),
+      first_timestamp_in_buffer_(0),
+      encoders_(new EncoderState[num_channels_]),
+      interleave_buffer_(new uint8_t[2 * num_channels_]) {
+  CHECK_EQ(config.frame_size_ms % 10, 0)
+      << "Frame size must be an integer multiple of 10 ms.";
+  const int samples_per_channel =
+      kSampleRateHz / 100 * num_10ms_frames_per_packet_;
+  for (int i = 0; i < num_channels_; ++i) {
+    encoders_[i].speech_buffer.reset(new int16_t[samples_per_channel]);
+    encoders_[i].encoded_buffer.reset(new uint8_t[samples_per_channel / 2]);
+  }
+}
+
+AudioEncoderG722::~AudioEncoderG722() {}
+
+int AudioEncoderG722::sample_rate_hz() const {
+  return kSampleRateHz;
+}
+int AudioEncoderG722::num_channels() const {
+  return num_channels_;
+}
+int AudioEncoderG722::Num10MsFramesInNextPacket() const {
+  return num_10ms_frames_per_packet_;
+}
+
+bool AudioEncoderG722::Encode(uint32_t timestamp,
+                              const int16_t* audio,
+                              size_t max_encoded_bytes,
+                              uint8_t* encoded,
+                              size_t* encoded_bytes,
+                              EncodedInfo* info) {
+  const int samples_per_channel =
+      kSampleRateHz / 100 * num_10ms_frames_per_packet_;
+  CHECK_GE(max_encoded_bytes,
+           static_cast<size_t>(samples_per_channel) / 2 * num_channels_);
+
+  if (num_10ms_frames_buffered_ == 0)
+    first_timestamp_in_buffer_ = timestamp;
+
+  // Deinterleave samples and save them in each channel's buffer.
+  const int start = kSampleRateHz / 100 * num_10ms_frames_buffered_;
+  for (int i = 0; i < kSampleRateHz / 100; ++i)
+    for (int j = 0; j < num_channels_; ++j)
+      encoders_[j].speech_buffer[start + i] = audio[i * num_channels_ + j];
+
+  // If we don't yet have enough samples for a packet, we're done for now.
+  if (++num_10ms_frames_buffered_ < num_10ms_frames_per_packet_) {
+    *encoded_bytes = 0;
+    return true;
+  }
+
+  // Encode each channel separately.
+  CHECK_EQ(num_10ms_frames_buffered_, num_10ms_frames_per_packet_);
+  num_10ms_frames_buffered_ = 0;
+  for (int i = 0; i < num_channels_; ++i) {
+    const int encoded = WebRtcG722_Encode(
+        encoders_[i].encoder, encoders_[i].speech_buffer.get(),
+        samples_per_channel, encoders_[i].encoded_buffer.get());
+    if (encoded < 0)
+      return false;
+    CHECK_EQ(encoded, samples_per_channel / 2);
+  }
+
+  // Interleave the encoded bytes of the different channels. Each separate
+  // channel and the interleaved stream encodes two samples per byte, most
+  // significant half first.
+  for (int i = 0; i < samples_per_channel / 2; ++i) {
+    for (int j = 0; j < num_channels_; ++j) {
+      uint8_t two_samples = encoders_[j].encoded_buffer[i];
+      interleave_buffer_[j] = two_samples >> 4;
+      interleave_buffer_[num_channels_ + j] = two_samples & 0xf;
+    }
+    for (int j = 0; j < num_channels_; ++j)
+      encoded[i * num_channels_ + j] =
+          interleave_buffer_[2 * j] << 4 | interleave_buffer_[2 * j + 1];
+  }
+  *encoded_bytes = samples_per_channel / 2 * num_channels_;
+  info->encoded_timestamp = first_timestamp_in_buffer_;
+  return true;
+}
+
+}  // namespace webrtc

webrtc/modules/audio_coding/codecs/g722/g722.gypi

@@ -21,6 +21,8 @@
         ],
       },
       'sources': [
+        'audio_encoder_g722.cc',
+        'include/audio_encoder_g722.h',
         'include/g722_interface.h',
         'g722_interface.c',
         'g722_encode.c',

webrtc/modules/audio_coding/codecs/g722/g722_interface.c

@@ -46,9 +46,9 @@ int16_t WebRtcG722_FreeEncoder(G722EncInst *G722enc_inst)
 }
 
 int16_t WebRtcG722_Encode(G722EncInst *G722enc_inst,
-                          int16_t *speechIn,
+                          const int16_t* speechIn,
                           int16_t len,
-                          int16_t *encoded)
+                          uint8_t* encoded)
 {
     unsigned char *codechar = (unsigned char*) encoded;
     // Encode the input speech vector

webrtc/modules/audio_coding/codecs/g722/include/audio_encoder_g722.h

@@ -0,0 +1,64 @@
+/*
+ *  Copyright (c) 2014 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_G722_INCLUDE_AUDIO_ENCODER_G722_H_
+#define WEBRTC_MODULES_AUDIO_CODING_CODECS_G722_INCLUDE_AUDIO_ENCODER_G722_H_
+
+#include "webrtc/modules/audio_coding/codecs/audio_encoder.h"
+#include "webrtc/modules/audio_coding/codecs/g722/include/g722_interface.h"
+#include "webrtc/system_wrappers/interface/scoped_ptr.h"
+
+namespace webrtc {
+
+class AudioEncoderG722 : public AudioEncoder {
+ public:
+  struct Config {
+    Config() : payload_type(9), frame_size_ms(20), num_channels(1) {}
+
+    int payload_type;
+    int frame_size_ms;
+    int num_channels;
+  };
+
+  explicit AudioEncoderG722(const Config& config);
+  virtual ~AudioEncoderG722();
+
+  virtual int sample_rate_hz() const OVERRIDE;
+  virtual int num_channels() const OVERRIDE;
+  virtual int Num10MsFramesInNextPacket() const OVERRIDE;
+
+ protected:
+  virtual bool Encode(uint32_t timestamp,
+                      const int16_t* audio,
+                      size_t max_encoded_bytes,
+                      uint8_t* encoded,
+                      size_t* encoded_bytes,
+                      EncodedInfo* info) OVERRIDE;
+
+ private:
+  // The encoder state for one channel.
+  struct EncoderState {
+    G722EncInst* encoder;
+    scoped_ptr<int16_t[]> speech_buffer;  // Queued up for encoding.
+    scoped_ptr<uint8_t[]> encoded_buffer;  // Already encoded.
+    EncoderState();
+    ~EncoderState();
+  };
+
+  const int num_channels_;
+  const int num_10ms_frames_per_packet_;
+  int num_10ms_frames_buffered_;
+  uint32_t first_timestamp_in_buffer_;
+  const scoped_ptr<EncoderState[]> encoders_;
+  const scoped_ptr<uint8_t[]> interleave_buffer_;
+};
+
+}  // namespace webrtc
+#endif  // WEBRTC_MODULES_AUDIO_CODING_CODECS_G722_INCLUDE_AUDIO_ENCODER_G722_H_

webrtc/modules/audio_coding/codecs/g722/include/g722_interface.h

@@ -95,10 +95,10 @@ int16_t WebRtcG722_FreeEncoder(G722EncInst *G722enc_inst);
  *                              -1 - Error
  */
 
-int16_t WebRtcG722_Encode(G722EncInst *G722enc_inst,
-                          int16_t *speechIn,
+int16_t WebRtcG722_Encode(G722EncInst* G722enc_inst,
+                          const int16_t* speechIn,
                           int16_t len,
-                          int16_t *encoded);
+                          uint8_t* encoded);
 
 
 /****************************************************************************

webrtc/modules/audio_coding/codecs/g722/test/testG722.cc

@@ -62,7 +62,7 @@ int main(int argc, char* argv[])
     int16_t stream_len = 0;
     int16_t shortdata[960];
     int16_t decoded[960];
-    int16_t streamdata[80*3];
+    uint8_t streamdata[80 * 6];
     int16_t speechType[1];
 
     /* handling wrong input arguments in the command line */
@@ -124,7 +124,9 @@ int main(int argc, char* argv[])
 
         /* G.722 encoding + decoding */
         stream_len = WebRtcG722_Encode((G722EncInst *)G722enc_inst, shortdata, framelength, streamdata);
-        err = WebRtcG722_Decode((G722DecInst *)G722dec_inst, streamdata, stream_len, decoded, speechType);
+        err = WebRtcG722_Decode(G722dec_inst,
+                                reinterpret_cast<int16_t*>(streamdata),
+                                stream_len, decoded, speechType);
 
         /* Stop clock after call to encoder and decoder */
         runtime += (double)((clock()/(double)CLOCKS_PER_SEC_G722)-starttime);