Moving encoded_bytes into EncodedInfo

BUG=3926
R=kwiberg@webrtc.org

Review URL: https://webrtc-codereview.appspot.com/35469004

git-svn-id: http://webrtc.googlecode.com/svn/trunk@7883 4adac7df-926f-26a2-2b94-8c16560cd09d
This commit is contained in:
henrik.lundin@webrtc.org 2014-12-12 13:31:24 +00:00
parent c8bc717905
commit 3b79daff14
16 changed files with 53 additions and 69 deletions

View File

@ -23,6 +23,9 @@ namespace webrtc {
class AudioEncoder {
public:
struct EncodedInfo {
EncodedInfo() : encoded_bytes(0), encoded_timestamp(0), payload_type(0) {}
size_t encoded_bytes;
uint32_t encoded_timestamp;
int payload_type;
};
@ -41,7 +44,6 @@ class AudioEncoder {
size_t num_samples_per_channel,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) {
CHECK_EQ(num_samples_per_channel,
static_cast<size_t>(sample_rate_hz() / 100));
@ -49,9 +51,8 @@ class AudioEncoder {
audio,
max_encoded_bytes,
encoded,
encoded_bytes,
info);
CHECK_LE(*encoded_bytes, max_encoded_bytes);
CHECK_LE(info->encoded_bytes, max_encoded_bytes);
return ret;
}
@ -76,7 +77,6 @@ class AudioEncoder {
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) = 0;
};

View File

@ -90,13 +90,12 @@ bool AudioEncoderCng::EncodeInternal(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) {
DCHECK_GE(max_encoded_bytes, static_cast<size_t>(num_cng_coefficients_ + 1));
if (max_encoded_bytes < static_cast<size_t>(num_cng_coefficients_ + 1)) {
return false;
}
*encoded_bytes = 0;
info->encoded_bytes = 0;
const int num_samples = sample_rate_hz() / 100 * num_channels();
if (speech_buffer_.empty()) {
CHECK_EQ(frames_in_buffer_, 0);
@ -143,15 +142,14 @@ bool AudioEncoderCng::EncodeInternal(uint32_t timestamp,
bool return_val = true;
switch (activity) {
case Vad::kPassive: {
return_val = EncodePassive(encoded, encoded_bytes);
return_val = EncodePassive(encoded, &info->encoded_bytes);
info->encoded_timestamp = first_timestamp_in_buffer_;
info->payload_type = cng_payload_type_;
last_frame_active_ = false;
break;
}
case Vad::kActive: {
return_val =
EncodeActive(max_encoded_bytes, encoded, encoded_bytes, info);
return_val = EncodeActive(max_encoded_bytes, encoded, info);
last_frame_active_ = true;
break;
}
@ -190,17 +188,16 @@ bool AudioEncoderCng::EncodePassive(uint8_t* encoded, size_t* encoded_bytes) {
bool AudioEncoderCng::EncodeActive(size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) {
const size_t samples_per_10ms_frame = 10 * sample_rate_hz_ / 1000;
for (int i = 0; i < frames_in_buffer_; ++i) {
if (!speech_encoder_->Encode(first_timestamp_in_buffer_,
&speech_buffer_[i * samples_per_10ms_frame],
samples_per_10ms_frame, max_encoded_bytes,
encoded, encoded_bytes, info))
encoded, info))
return false;
if (i < frames_in_buffer_ - 1) {
CHECK_EQ(*encoded_bytes, 0u) << "Encoder delivered data too early.";
CHECK_EQ(info->encoded_bytes, 0u) << "Encoder delivered data too early.";
}
}
return true;

View File

@ -18,6 +18,7 @@ using ::testing::Return;
using ::testing::_;
using ::testing::SetArgPointee;
using ::testing::InSequence;
using ::testing::Invoke;
namespace webrtc {
@ -33,8 +34,7 @@ class AudioEncoderCngTest : public ::testing::Test {
AudioEncoderCngTest()
: mock_vad_(new MockVad(Vad::kVadNormal)),
timestamp_(4711),
num_audio_samples_10ms_(0),
encoded_bytes_(0) {
num_audio_samples_10ms_(0) {
memset(encoded_, 0, kMaxEncodedBytes);
memset(audio_, 0, kMaxNumSamples * 2);
config_.speech_encoder = &mock_encoder_;
@ -70,11 +70,9 @@ class AudioEncoderCngTest : public ::testing::Test {
void Encode() {
ASSERT_TRUE(cng_) << "Must call CreateCng() first.";
memset(&encoded_info_, 0, sizeof(encoded_info_));
encoded_bytes_ = 0;
encoded_info_ = AudioEncoder::EncodedInfo();
ASSERT_TRUE(cng_->Encode(timestamp_, audio_, num_audio_samples_10ms_,
kMaxEncodedBytes, encoded_, &encoded_bytes_,
&encoded_info_));
kMaxEncodedBytes, encoded_, &encoded_info_));
timestamp_ += num_audio_samples_10ms_;
}
@ -89,30 +87,31 @@ class AudioEncoderCngTest : public ::testing::Test {
.WillRepeatedly(Return(active_speech ? Vad::kActive : Vad::kPassive));
// Don't expect any calls to the encoder yet.
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _, _)).Times(0);
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _)).Times(0);
for (int i = 0; i < blocks_per_frame - 1; ++i) {
Encode();
EXPECT_EQ(0u, encoded_bytes_);
EXPECT_EQ(0u, encoded_info_.encoded_bytes);
}
AudioEncoder::EncodedInfo info;
if (active_speech) {
// Now expect |blocks_per_frame| calls to the encoder in sequence.
// Let the speech codec mock return true and set the number of encoded
// bytes to |kMockReturnEncodedBytes|.
InSequence s;
for (int j = 0; j < blocks_per_frame - 1; ++j) {
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _, _))
.WillOnce(DoAll(SetArgPointee<4>(0), Return(true)));
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.WillOnce(DoAll(SetArgPointee<4>(info), Return(true)));
}
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _, _))
.WillOnce(
DoAll(SetArgPointee<4>(kMockReturnEncodedBytes), Return(true)));
info.encoded_bytes = kMockReturnEncodedBytes;
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.WillOnce(DoAll(SetArgPointee<4>(info), Return(true)));
}
Encode();
if (active_speech) {
EXPECT_EQ(kMockReturnEncodedBytes, encoded_bytes_);
EXPECT_EQ(kMockReturnEncodedBytes, encoded_info_.encoded_bytes);
} else {
EXPECT_EQ(static_cast<size_t>(config_.num_cng_coefficients + 1),
encoded_bytes_);
encoded_info_.encoded_bytes);
}
}
@ -184,7 +183,6 @@ class AudioEncoderCngTest : public ::testing::Test {
int16_t audio_[kMaxNumSamples];
size_t num_audio_samples_10ms_;
uint8_t encoded_[kMaxEncodedBytes];
size_t encoded_bytes_;
AudioEncoder::EncodedInfo encoded_info_;
};
@ -239,7 +237,7 @@ TEST_F(AudioEncoderCngTest, EncodePassive) {
EXPECT_CALL(*mock_vad_, VoiceActivity(_, _, _))
.WillRepeatedly(Return(Vad::kPassive));
// Expect no calls at all to the speech encoder mock.
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _, _)).Times(0);
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _)).Times(0);
uint32_t expected_timestamp = timestamp_;
for (int i = 0; i < 100; ++i) {
Encode();
@ -251,13 +249,13 @@ TEST_F(AudioEncoderCngTest, EncodePassive) {
// If so, verify that we got a CNG encoding.
EXPECT_EQ(kCngPayloadType, encoded_info_.payload_type);
EXPECT_EQ(static_cast<size_t>(config_.num_cng_coefficients) + 1,
encoded_bytes_);
encoded_info_.encoded_bytes);
EXPECT_EQ(expected_timestamp, encoded_info_.encoded_timestamp);
}
expected_timestamp += kBlocksPerFrame * num_audio_samples_10ms_;
} else {
// Otherwise, expect no output.
EXPECT_EQ(0u, encoded_bytes_);
EXPECT_EQ(0u, encoded_info_.encoded_bytes);
}
}
}
@ -268,19 +266,19 @@ TEST_F(AudioEncoderCngTest, MixedActivePassive) {
CreateCng();
// All of the frame is active speech.
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _, _))
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.Times(6)
.WillRepeatedly(Return(true));
EXPECT_TRUE(CheckMixedActivePassive(Vad::kActive, Vad::kActive));
// First half of the frame is active speech.
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _, _))
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.Times(6)
.WillRepeatedly(Return(true));
EXPECT_TRUE(CheckMixedActivePassive(Vad::kActive, Vad::kPassive));
// Second half of the frame is active speech.
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _, _))
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.Times(6)
.WillRepeatedly(Return(true));
EXPECT_TRUE(CheckMixedActivePassive(Vad::kPassive, Vad::kActive));
@ -324,7 +322,7 @@ TEST_F(AudioEncoderCngTest, VadInputSize60Ms) {
// speech encoder.
TEST_F(AudioEncoderCngTest, VerifyEncoderInfoPropagation) {
CreateCng();
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _, &encoded_info_))
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, &encoded_info_))
.WillOnce(Return(true));
EXPECT_CALL(mock_encoder_, Num10MsFramesInNextPacket()).WillOnce(Return(1));
EXPECT_CALL(*mock_vad_, VoiceActivity(_, _, _))
@ -335,7 +333,7 @@ TEST_F(AudioEncoderCngTest, VerifyEncoderInfoPropagation) {
// Verifies that the correct payload type is set when CNG is encoded.
TEST_F(AudioEncoderCngTest, VerifyCngPayloadType) {
CreateCng();
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _, _)).Times(0);
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _)).Times(0);
EXPECT_CALL(mock_encoder_, Num10MsFramesInNextPacket()).WillOnce(Return(1));
EXPECT_CALL(*mock_vad_, VoiceActivity(_, _, _))
.WillOnce(Return(Vad::kPassive));
@ -357,20 +355,22 @@ TEST_F(AudioEncoderCngTest, VerifySidFrameAfterSpeech) {
Encode();
EXPECT_EQ(kCngPayloadType, encoded_info_.payload_type);
EXPECT_EQ(static_cast<size_t>(config_.num_cng_coefficients) + 1,
encoded_bytes_);
encoded_info_.encoded_bytes);
// Encode again, and make sure we got no frame at all (since the SID frame
// period is 100 ms by default).
Encode();
EXPECT_EQ(0u, encoded_bytes_);
EXPECT_EQ(0u, encoded_info_.encoded_bytes);
// Now encode active speech.
encoded_info_.payload_type = 0;
EXPECT_CALL(*mock_vad_, VoiceActivity(_, _, _))
.WillOnce(Return(Vad::kActive));
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _, _))
.WillOnce(DoAll(SetArgPointee<4>(kMockReturnEncodedBytes), Return(true)));
AudioEncoder::EncodedInfo info;
info.encoded_bytes = kMockReturnEncodedBytes;
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.WillOnce(DoAll(SetArgPointee<4>(info), Return(true)));
Encode();
EXPECT_EQ(kMockReturnEncodedBytes, encoded_bytes_);
EXPECT_EQ(kMockReturnEncodedBytes, encoded_info_.encoded_bytes);
// Go back to noise again, and verify that a SID frame is emitted.
EXPECT_CALL(*mock_vad_, VoiceActivity(_, _, _))
@ -378,7 +378,7 @@ TEST_F(AudioEncoderCngTest, VerifySidFrameAfterSpeech) {
Encode();
EXPECT_EQ(kCngPayloadType, encoded_info_.payload_type);
EXPECT_EQ(static_cast<size_t>(config_.num_cng_coefficients) + 1,
encoded_bytes_);
encoded_info_.encoded_bytes);
}
#if GTEST_HAS_DEATH_TEST && !defined(WEBRTC_ANDROID)

View File

@ -57,7 +57,6 @@ class AudioEncoderCng : public AudioEncoder {
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) OVERRIDE;
private:
@ -71,7 +70,6 @@ class AudioEncoderCng : public AudioEncoder {
bool EncodeActive(size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info);
AudioEncoder* speech_encoder_;

View File

@ -63,7 +63,6 @@ bool AudioEncoderPcm::EncodeInternal(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) {
const int num_samples = sample_rate_hz() / 100 * num_channels();
if (speech_buffer_.empty()) {
@ -73,7 +72,7 @@ bool AudioEncoderPcm::EncodeInternal(uint32_t timestamp,
speech_buffer_.push_back(audio[i]);
}
if (speech_buffer_.size() < static_cast<size_t>(full_frame_samples_)) {
*encoded_bytes = 0;
info->encoded_bytes = 0;
return true;
}
CHECK_EQ(speech_buffer_.size(), static_cast<size_t>(full_frame_samples_));
@ -83,7 +82,7 @@ bool AudioEncoderPcm::EncodeInternal(uint32_t timestamp,
info->payload_type = payload_type_;
if (ret < 0)
return false;
*encoded_bytes = static_cast<size_t>(ret);
info->encoded_bytes = static_cast<size_t>(ret);
return true;
}

View File

@ -44,7 +44,6 @@ class AudioEncoderPcm : public AudioEncoder {
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) OVERRIDE;
virtual int16_t EncodeCall(const int16_t* audio,

View File

@ -68,7 +68,6 @@ bool AudioEncoderG722::EncodeInternal(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_;
@ -86,7 +85,7 @@ bool AudioEncoderG722::EncodeInternal(uint32_t timestamp,
// 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;
info->encoded_bytes = 0;
return true;
}
@ -115,7 +114,7 @@ bool AudioEncoderG722::EncodeInternal(uint32_t timestamp,
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_bytes = samples_per_channel / 2 * num_channels_;
info->encoded_timestamp = first_timestamp_in_buffer_;
info->payload_type = payload_type_;
return true;

View File

@ -40,7 +40,6 @@ class AudioEncoderG722 : public AudioEncoder {
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) OVERRIDE;
private:

View File

@ -56,7 +56,6 @@ bool AudioEncoderIlbc::EncodeInternal(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) {
const size_t expected_output_len =
num_10ms_frames_per_packet_ == 2 ? 38 : 50;
@ -74,7 +73,7 @@ bool AudioEncoderIlbc::EncodeInternal(uint32_t timestamp,
// If we don't yet have enough buffered input for a whole packet, we're done
// for now.
if (++num_10ms_frames_buffered_ < num_10ms_frames_per_packet_) {
*encoded_bytes = 0;
info->encoded_bytes = 0;
return true;
}
@ -89,7 +88,7 @@ bool AudioEncoderIlbc::EncodeInternal(uint32_t timestamp,
if (output_len == -1)
return false; // Encoding error.
DCHECK_EQ(output_len, static_cast<int>(expected_output_len));
*encoded_bytes = output_len;
info->encoded_bytes = output_len;
info->encoded_timestamp = first_timestamp_in_buffer_;
info->payload_type = payload_type_;
return true;

View File

@ -39,7 +39,6 @@ class AudioEncoderIlbc : public AudioEncoder {
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) OVERRIDE;
private:

View File

@ -87,7 +87,6 @@ class AudioEncoderDecoderIsac : public AudioEncoder, public AudioDecoder {
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) OVERRIDE;
private:

View File

@ -111,7 +111,6 @@ bool AudioEncoderDecoderIsac::EncodeInternal(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) {
if (!packet_in_progress_) {
// Starting a new packet; remember the timestamp for later.
@ -133,7 +132,7 @@ bool AudioEncoderDecoderIsac::EncodeInternal(uint32_t timestamp,
// buffer. All we can do is check for an overrun after the fact.
CHECK(static_cast<size_t>(r) <= max_encoded_bytes);
*encoded_bytes = r;
info->encoded_bytes = r;
if (r > 0) {
// Got enough input to produce a packet. Return the saved timestamp from
// the first chunk of input that went into the packet.

View File

@ -26,12 +26,11 @@ class MockAudioEncoder : public AudioEncoder {
MOCK_CONST_METHOD0(Num10MsFramesInNextPacket, int());
MOCK_CONST_METHOD0(Max10MsFramesInAPacket, int());
// Note, we explicitly chose not to create a mock for the Encode method.
MOCK_METHOD6(EncodeInternal,
MOCK_METHOD5(EncodeInternal,
bool(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info));
};

View File

@ -82,7 +82,6 @@ bool AudioEncoderOpus::EncodeInternal(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) {
if (input_buffer_.empty())
first_timestamp_in_buffer_ = timestamp;
@ -90,7 +89,7 @@ bool AudioEncoderOpus::EncodeInternal(uint32_t timestamp,
audio + samples_per_10ms_frame_);
if (input_buffer_.size() < (static_cast<size_t>(num_10ms_frames_per_packet_) *
samples_per_10ms_frame_)) {
*encoded_bytes = 0;
info->encoded_bytes = 0;
return true;
}
CHECK_EQ(input_buffer_.size(),
@ -103,7 +102,7 @@ bool AudioEncoderOpus::EncodeInternal(uint32_t timestamp,
input_buffer_.clear();
if (r < 0)
return false;
*encoded_bytes = r;
info->encoded_bytes = r;
info->encoded_timestamp = first_timestamp_in_buffer_;
info->payload_type = payload_type_;
return true;

View File

@ -41,7 +41,6 @@ class AudioEncoderOpus : public AudioEncoder {
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
EncodedInfo* info) OVERRIDE;
private:

View File

@ -135,14 +135,14 @@ class AudioDecoderTest : public ::testing::Test {
virtual int EncodeFrame(const int16_t* input,
size_t input_len_samples,
uint8_t* output) {
size_t enc_len_bytes = 0;
encoded_info_.encoded_bytes = 0;
const size_t samples_per_10ms = audio_encoder_->sample_rate_hz() / 100;
CHECK_EQ(samples_per_10ms * audio_encoder_->Num10MsFramesInNextPacket(),
input_len_samples);
scoped_ptr<int16_t[]> interleaved_input(
new int16_t[channels_ * samples_per_10ms]);
for (int i = 0; i < audio_encoder_->Num10MsFramesInNextPacket(); ++i) {
EXPECT_EQ(0u, enc_len_bytes);
EXPECT_EQ(0u, encoded_info_.encoded_bytes);
// Duplicate the mono input signal to however many channels the test
// wants.
@ -152,10 +152,10 @@ class AudioDecoderTest : public ::testing::Test {
EXPECT_TRUE(audio_encoder_->Encode(
0, interleaved_input.get(), audio_encoder_->sample_rate_hz() / 100,
data_length_ * 2, output, &enc_len_bytes, &encoded_info_));
data_length_ * 2, output, &encoded_info_));
}
EXPECT_EQ(payload_type_, encoded_info_.payload_type);
return static_cast<int>(enc_len_bytes);
return static_cast<int>(encoded_info_.encoded_bytes);
}
// Encodes and decodes audio. The absolute difference between the input and