Reland 8749: AudioEncoder: return EncodedInfo from Encode() and EncodeInternal()

Old review at:
https://webrtc-codereview.appspot.com/43839004/

R=kwiberg@webrtc.org

Review URL: https://webrtc-codereview.appspot.com/45769004

Cr-Commit-Position: refs/heads/master@{#8788}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8788 4adac7df-926f-26a2-2b94-8c16560cd09d
This commit is contained in:
jmarusic@webrtc.org 2015-03-19 08:50:26 +00:00
parent d21406d333
commit 9afaee74ab
22 changed files with 219 additions and 236 deletions

View File

@ -19,16 +19,17 @@ AudioEncoder::EncodedInfo::EncodedInfo() : EncodedInfoLeaf() {
AudioEncoder::EncodedInfo::~EncodedInfo() {
}
void AudioEncoder::Encode(uint32_t rtp_timestamp,
const int16_t* audio,
size_t num_samples_per_channel,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) {
AudioEncoder::EncodedInfo AudioEncoder::Encode(uint32_t rtp_timestamp,
const int16_t* audio,
size_t num_samples_per_channel,
size_t max_encoded_bytes,
uint8_t* encoded) {
CHECK_EQ(num_samples_per_channel,
static_cast<size_t>(SampleRateHz() / 100));
EncodeInternal(rtp_timestamp, audio, max_encoded_bytes, encoded, info);
CHECK_LE(info->encoded_bytes, max_encoded_bytes);
EncodedInfo info =
EncodeInternal(rtp_timestamp, audio, max_encoded_bytes, encoded);
CHECK_LE(info.encoded_bytes, max_encoded_bytes);
return info;
}
int AudioEncoder::RtpTimestampRateHz() const {

View File

@ -58,16 +58,15 @@ class AudioEncoder {
// Accepts one 10 ms block of input audio (i.e., sample_rate_hz() / 100 *
// num_channels() samples). Multi-channel audio must be sample-interleaved.
// The encoder produces zero or more bytes of output in |encoded|,
// and provides additional encoding information in |info|.
// The encoder produces zero or more bytes of output in |encoded| and
// returns additional encoding information.
// The caller is responsible for making sure that |max_encoded_bytes| is
// not smaller than the number of bytes actually produced by the encoder.
void Encode(uint32_t rtp_timestamp,
const int16_t* audio,
size_t num_samples_per_channel,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info);
EncodedInfo Encode(uint32_t rtp_timestamp,
const int16_t* audio,
size_t num_samples_per_channel,
size_t max_encoded_bytes,
uint8_t* encoded);
// Return the input sample rate in Hz and the number of input channels.
// These are constants set at instantiation time.
@ -107,11 +106,10 @@ class AudioEncoder {
virtual void SetProjectedPacketLossRate(double fraction) {}
protected:
virtual void EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) = 0;
virtual EncodedInfo EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) = 0;
};
} // namespace webrtc

View File

@ -109,13 +109,12 @@ void AudioEncoderCng::SetProjectedPacketLossRate(double fraction) {
speech_encoder_->SetProjectedPacketLossRate(fraction);
}
void AudioEncoderCng::EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) {
AudioEncoder::EncodedInfo AudioEncoderCng::EncodeInternal(
uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) {
CHECK_GE(max_encoded_bytes, static_cast<size_t>(num_cng_coefficients_ + 1));
info->encoded_bytes = 0;
const int num_samples = SampleRateHz() / 100 * NumChannels();
if (speech_buffer_.empty()) {
CHECK_EQ(frames_in_buffer_, 0);
@ -126,7 +125,7 @@ void AudioEncoderCng::EncodeInternal(uint32_t rtp_timestamp,
}
++frames_in_buffer_;
if (frames_in_buffer_ < speech_encoder_->Num10MsFramesInNextPacket()) {
return;
return EncodedInfo();
}
CHECK_LE(frames_in_buffer_ * 10, kMaxFrameSizeMs)
<< "Frame size cannot be larger than " << kMaxFrameSizeMs
@ -159,14 +158,15 @@ void AudioEncoderCng::EncodeInternal(uint32_t rtp_timestamp,
samples_per_10ms_frame * blocks_in_second_vad_call, SampleRateHz());
}
EncodedInfo info;
switch (activity) {
case Vad::kPassive: {
EncodePassive(max_encoded_bytes, encoded, info);
info = EncodePassive(max_encoded_bytes, encoded);
last_frame_active_ = false;
break;
}
case Vad::kActive: {
EncodeActive(max_encoded_bytes, encoded, info);
info = EncodeActive(max_encoded_bytes, encoded);
last_frame_active_ = true;
break;
}
@ -178,15 +178,17 @@ void AudioEncoderCng::EncodeInternal(uint32_t rtp_timestamp,
speech_buffer_.clear();
frames_in_buffer_ = 0;
return info;
}
void AudioEncoderCng::EncodePassive(size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) {
AudioEncoder::EncodedInfo AudioEncoderCng::EncodePassive(
size_t max_encoded_bytes,
uint8_t* encoded) {
bool force_sid = last_frame_active_;
bool output_produced = false;
const size_t samples_per_10ms_frame = SamplesPer10msFrame();
CHECK_GE(max_encoded_bytes, frames_in_buffer_ * samples_per_10ms_frame);
AudioEncoder::EncodedInfo info;
for (int i = 0; i < frames_in_buffer_; ++i) {
int16_t encoded_bytes_tmp = 0;
CHECK_GE(WebRtcCng_Encode(cng_inst_.get(),
@ -195,30 +197,32 @@ void AudioEncoderCng::EncodePassive(size_t max_encoded_bytes,
encoded, &encoded_bytes_tmp, force_sid), 0);
if (encoded_bytes_tmp > 0) {
CHECK(!output_produced);
info->encoded_bytes = static_cast<size_t>(encoded_bytes_tmp);
info.encoded_bytes = static_cast<size_t>(encoded_bytes_tmp);
output_produced = true;
force_sid = false;
}
}
info->encoded_timestamp = first_timestamp_in_buffer_;
info->payload_type = cng_payload_type_;
info->send_even_if_empty = true;
info->speech = false;
info.encoded_timestamp = first_timestamp_in_buffer_;
info.payload_type = cng_payload_type_;
info.send_even_if_empty = true;
info.speech = false;
return info;
}
void AudioEncoderCng::EncodeActive(size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) {
AudioEncoder::EncodedInfo AudioEncoderCng::EncodeActive(
size_t max_encoded_bytes,
uint8_t* encoded) {
const size_t samples_per_10ms_frame = SamplesPer10msFrame();
AudioEncoder::EncodedInfo info;
for (int i = 0; i < frames_in_buffer_; ++i) {
speech_encoder_->Encode(first_timestamp_in_buffer_,
&speech_buffer_[i * samples_per_10ms_frame],
samples_per_10ms_frame, max_encoded_bytes,
encoded, info);
info = speech_encoder_->Encode(
first_timestamp_in_buffer_, &speech_buffer_[i * samples_per_10ms_frame],
samples_per_10ms_frame, max_encoded_bytes, encoded);
if (i < frames_in_buffer_ - 1) {
CHECK_EQ(info->encoded_bytes, 0u) << "Encoder delivered data too early.";
CHECK_EQ(info.encoded_bytes, 0u) << "Encoder delivered data too early.";
}
}
return info;
}
size_t AudioEncoderCng::SamplesPer10msFrame() const {

View File

@ -75,9 +75,8 @@ class AudioEncoderCngTest : public ::testing::Test {
void Encode() {
ASSERT_TRUE(cng_) << "Must call CreateCng() first.";
encoded_info_ = AudioEncoder::EncodedInfo();
cng_->Encode(timestamp_, audio_, num_audio_samples_10ms_,
encoded_.size(), &encoded_[0], &encoded_info_);
encoded_info_ = cng_->Encode(timestamp_, audio_, num_audio_samples_10ms_,
encoded_.size(), &encoded_[0]);
timestamp_ += num_audio_samples_10ms_;
}
@ -92,24 +91,24 @@ 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_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;
AudioEncoder::EncodedInfo info;
for (int j = 0; j < blocks_per_frame - 1; ++j) {
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.WillOnce(SetArgPointee<4>(info));
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.WillOnce(Return(info));
}
info.encoded_bytes = kMockReturnEncodedBytes;
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.WillOnce(SetArgPointee<4>(info));
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.WillOnce(Return(info));
}
Encode();
if (active_speech) {
@ -254,7 +253,7 @@ 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();
@ -284,20 +283,23 @@ TEST_F(AudioEncoderCngTest, MixedActivePassive) {
CreateCng();
// All of the frame is active speech.
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.Times(6);
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.Times(6)
.WillRepeatedly(Return(AudioEncoder::EncodedInfo()));
EXPECT_TRUE(CheckMixedActivePassive(Vad::kActive, Vad::kActive));
EXPECT_TRUE(encoded_info_.speech);
// First half of the frame is active speech.
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.Times(6);
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.Times(6)
.WillRepeatedly(Return(AudioEncoder::EncodedInfo()));
EXPECT_TRUE(CheckMixedActivePassive(Vad::kActive, Vad::kPassive));
EXPECT_TRUE(encoded_info_.speech);
// Second half of the frame is active speech.
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.Times(6);
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.Times(6)
.WillRepeatedly(Return(AudioEncoder::EncodedInfo()));
EXPECT_TRUE(CheckMixedActivePassive(Vad::kPassive, Vad::kActive));
EXPECT_TRUE(encoded_info_.speech);
@ -336,22 +338,10 @@ TEST_F(AudioEncoderCngTest, VadInputSize60Ms) {
CheckVadInputSize(60, 30, 30);
}
// Verifies that the EncodedInfo struct pointer passed to
// AudioEncoderCng::Encode is propagated to the Encode call to the underlying
// speech encoder.
TEST_F(AudioEncoderCngTest, VerifyEncoderInfoPropagation) {
CreateCng();
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, &encoded_info_));
EXPECT_CALL(mock_encoder_, Num10MsFramesInNextPacket()).WillOnce(Return(1));
EXPECT_CALL(*mock_vad_, VoiceActivity(_, _, _))
.WillOnce(Return(Vad::kActive));
Encode();
}
// 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));
@ -385,8 +375,7 @@ TEST_F(AudioEncoderCngTest, VerifySidFrameAfterSpeech) {
.WillOnce(Return(Vad::kActive));
AudioEncoder::EncodedInfo info;
info.encoded_bytes = kMockReturnEncodedBytes;
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.WillOnce(SetArgPointee<4>(info));
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _)).WillOnce(Return(info));
Encode();
EXPECT_EQ(kMockReturnEncodedBytes, encoded_info_.encoded_bytes);

View File

@ -56,11 +56,10 @@ class AudioEncoderCng final : public AudioEncoder {
void SetProjectedPacketLossRate(double fraction) override;
protected:
void EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) override;
EncodedInfo EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) override;
private:
// Deleter for use with scoped_ptr. E.g., use as
@ -69,12 +68,8 @@ class AudioEncoderCng final : public AudioEncoder {
inline void operator()(CNG_enc_inst* ptr) const { WebRtcCng_FreeEnc(ptr); }
};
void EncodePassive(size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info);
void EncodeActive(size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info);
EncodedInfo EncodePassive(size_t max_encoded_bytes, uint8_t* encoded);
EncodedInfo EncodeActive(size_t max_encoded_bytes, uint8_t* encoded);
size_t SamplesPer10msFrame() const;
AudioEncoder* speech_encoder_;

View File

@ -66,11 +66,11 @@ int AudioEncoderPcm::Max10MsFramesInAPacket() const {
return num_10ms_frames_per_packet_;
}
void AudioEncoderPcm::EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) {
AudioEncoder::EncodedInfo AudioEncoderPcm::EncodeInternal(
uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) {
const int num_samples = SampleRateHz() / 100 * NumChannels();
if (speech_buffer_.empty()) {
first_timestamp_in_buffer_ = rtp_timestamp;
@ -79,17 +79,18 @@ void AudioEncoderPcm::EncodeInternal(uint32_t rtp_timestamp,
speech_buffer_.push_back(audio[i]);
}
if (speech_buffer_.size() < full_frame_samples_) {
info->encoded_bytes = 0;
return;
return EncodedInfo();
}
CHECK_EQ(speech_buffer_.size(), full_frame_samples_);
CHECK_GE(max_encoded_bytes, full_frame_samples_);
int16_t ret = EncodeCall(&speech_buffer_[0], full_frame_samples_, encoded);
CHECK_GE(ret, 0);
speech_buffer_.clear();
info->encoded_timestamp = first_timestamp_in_buffer_;
info->payload_type = payload_type_;
info->encoded_bytes = static_cast<size_t>(ret);
EncodedInfo info;
info.encoded_timestamp = first_timestamp_in_buffer_;
info.payload_type = payload_type_;
info.encoded_bytes = static_cast<size_t>(ret);
return info;
}
int16_t AudioEncoderPcmA::EncodeCall(const int16_t* audio,

View File

@ -41,11 +41,10 @@ class AudioEncoderPcm : public AudioEncoder {
protected:
AudioEncoderPcm(const Config& config, int sample_rate_hz);
void EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) override;
EncodedInfo EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) override;
virtual int16_t EncodeCall(const int16_t* audio,
size_t input_len,

View File

@ -77,11 +77,11 @@ int AudioEncoderG722::Max10MsFramesInAPacket() const {
return num_10ms_frames_per_packet_;
}
void AudioEncoderG722::EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) {
AudioEncoder::EncodedInfo AudioEncoderG722::EncodeInternal(
uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) {
CHECK_GE(max_encoded_bytes, MaxEncodedBytes());
if (num_10ms_frames_buffered_ == 0)
@ -95,8 +95,7 @@ void AudioEncoderG722::EncodeInternal(uint32_t rtp_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_) {
info->encoded_bytes = 0;
return;
return EncodedInfo();
}
// Encode each channel separately.
@ -124,9 +123,11 @@ void AudioEncoderG722::EncodeInternal(uint32_t rtp_timestamp,
encoded[i * num_channels_ + j] =
interleave_buffer_[2 * j] << 4 | interleave_buffer_[2 * j + 1];
}
info->encoded_bytes = samples_per_channel / 2 * num_channels_;
info->encoded_timestamp = first_timestamp_in_buffer_;
info->payload_type = payload_type_;
EncodedInfo info;
info.encoded_bytes = samples_per_channel / 2 * num_channels_;
info.encoded_timestamp = first_timestamp_in_buffer_;
info.payload_type = payload_type_;
return info;
}
int AudioEncoderG722::SamplesPerChannel() const {

View File

@ -38,11 +38,10 @@ class AudioEncoderG722 : public AudioEncoder {
int Max10MsFramesInAPacket() const override;
protected:
void EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) override;
EncodedInfo EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) override;
private:
// The encoder state for one channel.

View File

@ -63,11 +63,11 @@ int AudioEncoderIlbc::Max10MsFramesInAPacket() const {
return num_10ms_frames_per_packet_;
}
void AudioEncoderIlbc::EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) {
AudioEncoder::EncodedInfo AudioEncoderIlbc::EncodeInternal(
uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) {
DCHECK_GE(max_encoded_bytes, RequiredOutputSizeBytes());
// Save timestamp if starting a new packet.
@ -82,8 +82,7 @@ void AudioEncoderIlbc::EncodeInternal(uint32_t rtp_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_) {
info->encoded_bytes = 0;
return;
return EncodedInfo();
}
// Encode buffered input.
@ -95,10 +94,12 @@ void AudioEncoderIlbc::EncodeInternal(uint32_t rtp_timestamp,
kSampleRateHz / 100 * num_10ms_frames_per_packet_,
encoded);
CHECK_GE(output_len, 0);
info->encoded_bytes = output_len;
DCHECK_EQ(info->encoded_bytes, RequiredOutputSizeBytes());
info->encoded_timestamp = first_timestamp_in_buffer_;
info->payload_type = payload_type_;
EncodedInfo info;
info.encoded_bytes = output_len;
DCHECK_EQ(info.encoded_bytes, RequiredOutputSizeBytes());
info.encoded_timestamp = first_timestamp_in_buffer_;
info.payload_type = payload_type_;
return info;
}
size_t AudioEncoderIlbc::RequiredOutputSizeBytes() const {

View File

@ -38,11 +38,10 @@ class AudioEncoderIlbc : public AudioEncoder {
int Max10MsFramesInAPacket() const override;
protected:
void EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) override;
EncodedInfo EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) override;
private:
size_t RequiredOutputSizeBytes() const;

View File

@ -85,11 +85,10 @@ class AudioEncoderDecoderIsacT : public AudioEncoder, public AudioDecoder {
protected:
// AudioEncoder protected method.
void EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) override;
EncodedInfo EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) override;
// AudioDecoder protected method.
int DecodeInternal(const uint8_t* encoded,

View File

@ -184,11 +184,11 @@ int AudioEncoderDecoderIsacT<T>::Max10MsFramesInAPacket() const {
}
template <typename T>
void AudioEncoderDecoderIsacT<T>::EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) {
AudioEncoder::EncodedInfo AudioEncoderDecoderIsacT<T>::EncodeInternal(
uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) {
CriticalSectionScoped cs_lock(lock_.get());
if (!packet_in_progress_) {
// Starting a new packet; remember the timestamp for later.
@ -206,15 +206,17 @@ void AudioEncoderDecoderIsacT<T>::EncodeInternal(uint32_t rtp_timestamp,
// buffer. All we can do is check for an overrun after the fact.
CHECK(static_cast<size_t>(r) <= max_encoded_bytes);
info->encoded_bytes = r;
if (r == 0)
return;
return EncodedInfo();
// Got enough input to produce a packet. Return the saved timestamp from
// the first chunk of input that went into the packet.
packet_in_progress_ = false;
info->encoded_timestamp = packet_timestamp_;
info->payload_type = payload_type_;
EncodedInfo info;
info.encoded_bytes = r;
info.encoded_timestamp = packet_timestamp_;
info.payload_type = payload_type_;
return info;
}
template <typename T>

View File

@ -29,12 +29,11 @@ class MockAudioEncoder : public AudioEncoder {
MOCK_METHOD1(SetTargetBitrate, void(int));
MOCK_METHOD1(SetProjectedPacketLossRate, void(double));
// Note, we explicitly chose not to create a mock for the Encode method.
MOCK_METHOD5(EncodeInternal,
void(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info));
MOCK_METHOD4(EncodeInternal,
EncodedInfo(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded));
};
} // namespace webrtc

View File

@ -183,19 +183,18 @@ void AudioEncoderOpus::SetProjectedPacketLossRate(double fraction) {
}
}
void AudioEncoderOpus::EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) {
AudioEncoder::EncodedInfo AudioEncoderOpus::EncodeInternal(
uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) {
if (input_buffer_.empty())
first_timestamp_in_buffer_ = rtp_timestamp;
input_buffer_.insert(input_buffer_.end(), audio,
audio + samples_per_10ms_frame_);
if (input_buffer_.size() < (static_cast<size_t>(num_10ms_frames_per_packet_) *
samples_per_10ms_frame_)) {
info->encoded_bytes = 0;
return;
return EncodedInfo();
}
CHECK_EQ(input_buffer_.size(),
static_cast<size_t>(num_10ms_frames_per_packet_) *
@ -207,12 +206,13 @@ void AudioEncoderOpus::EncodeInternal(uint32_t rtp_timestamp,
ClampInt16(max_encoded_bytes), encoded);
CHECK_GE(r, 0); // Fails only if fed invalid data.
input_buffer_.clear();
info->encoded_bytes = r;
info->encoded_timestamp = first_timestamp_in_buffer_;
info->payload_type = payload_type_;
// Allows Opus to send empty packets.
info->send_even_if_empty = true;
info->speech = r > 0;
EncodedInfo info;
info.encoded_bytes = r;
info.encoded_timestamp = first_timestamp_in_buffer_;
info.payload_type = payload_type_;
info.send_even_if_empty = true; // Allows Opus to send empty packets.
info.speech = r > 0;
return info;
}
} // namespace webrtc

View File

@ -58,11 +58,10 @@ class AudioEncoderOpus final : public AudioEncoder {
bool dtx_enabled() const { return dtx_enabled_; }
protected:
void EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) override;
EncodedInfo EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) override;
private:
const int num_10ms_frames_per_packet_;

View File

@ -60,48 +60,48 @@ void AudioEncoderCopyRed::SetProjectedPacketLossRate(double fraction) {
speech_encoder_->SetProjectedPacketLossRate(fraction);
}
void AudioEncoderCopyRed::EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) {
speech_encoder_->Encode(rtp_timestamp, audio,
static_cast<size_t>(SampleRateHz() / 100),
max_encoded_bytes, encoded, info);
AudioEncoder::EncodedInfo AudioEncoderCopyRed::EncodeInternal(
uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) {
EncodedInfo info = speech_encoder_->Encode(
rtp_timestamp, audio, static_cast<size_t>(SampleRateHz() / 100),
max_encoded_bytes, encoded);
CHECK_GE(max_encoded_bytes,
info->encoded_bytes + secondary_info_.encoded_bytes);
CHECK(info->redundant.empty()) << "Cannot use nested redundant encoders.";
info.encoded_bytes + secondary_info_.encoded_bytes);
CHECK(info.redundant.empty()) << "Cannot use nested redundant encoders.";
if (info->encoded_bytes > 0) {
if (info.encoded_bytes > 0) {
// |info| will be implicitly cast to an EncodedInfoLeaf struct, effectively
// discarding the (empty) vector of redundant information. This is
// intentional.
info->redundant.push_back(*info);
DCHECK_EQ(info->redundant.size(), 1u);
info.redundant.push_back(info);
DCHECK_EQ(info.redundant.size(), 1u);
if (secondary_info_.encoded_bytes > 0) {
memcpy(&encoded[info->encoded_bytes], secondary_encoded_.get(),
memcpy(&encoded[info.encoded_bytes], secondary_encoded_.get(),
secondary_info_.encoded_bytes);
info->redundant.push_back(secondary_info_);
DCHECK_EQ(info->redundant.size(), 2u);
info.redundant.push_back(secondary_info_);
DCHECK_EQ(info.redundant.size(), 2u);
}
// Save primary to secondary.
if (secondary_allocated_ < info->encoded_bytes) {
secondary_encoded_.reset(new uint8_t[info->encoded_bytes]);
secondary_allocated_ = info->encoded_bytes;
if (secondary_allocated_ < info.encoded_bytes) {
secondary_encoded_.reset(new uint8_t[info.encoded_bytes]);
secondary_allocated_ = info.encoded_bytes;
}
CHECK(secondary_encoded_);
memcpy(secondary_encoded_.get(), encoded, info->encoded_bytes);
secondary_info_ = *info;
DCHECK_EQ(info->speech, info->redundant[0].speech);
memcpy(secondary_encoded_.get(), encoded, info.encoded_bytes);
secondary_info_ = info;
DCHECK_EQ(info.speech, info.redundant[0].speech);
}
// Update main EncodedInfo.
info->payload_type = red_payload_type_;
info->encoded_bytes = 0;
for (std::vector<EncodedInfoLeaf>::const_iterator it =
info->redundant.begin();
it != info->redundant.end(); ++it) {
info->encoded_bytes += it->encoded_bytes;
info.payload_type = red_payload_type_;
info.encoded_bytes = 0;
for (std::vector<EncodedInfoLeaf>::const_iterator it = info.redundant.begin();
it != info.redundant.end(); ++it) {
info.encoded_bytes += it->encoded_bytes;
}
return info;
}
} // namespace webrtc

View File

@ -45,11 +45,10 @@ class AudioEncoderCopyRed : public AudioEncoder {
void SetProjectedPacketLossRate(double fraction) override;
protected:
void EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
EncodedInfo* info) override;
EncodedInfo EncodeInternal(uint32_t rtp_timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) override;
private:
AudioEncoder* speech_encoder_;

View File

@ -60,9 +60,8 @@ class AudioEncoderCopyRedTest : public ::testing::Test {
void Encode() {
ASSERT_TRUE(red_.get() != NULL);
encoded_info_ = AudioEncoder::EncodedInfo();
red_->Encode(timestamp_, audio_, num_audio_samples_10ms,
encoded_.size(), &encoded_[0], &encoded_info_);
encoded_info_ = red_->Encode(timestamp_, audio_, num_audio_samples_10ms,
encoded_.size(), &encoded_[0]);
timestamp_ += num_audio_samples_10ms;
}
@ -83,18 +82,16 @@ class MockEncodeHelper {
memset(&info_, 0, sizeof(info_));
}
void Encode(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
AudioEncoder::EncodedInfo* info) {
AudioEncoder::EncodedInfo Encode(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded) {
if (write_payload_) {
CHECK(encoded);
CHECK_LE(info_.encoded_bytes, max_encoded_bytes);
memcpy(encoded, payload_, info_.encoded_bytes);
}
CHECK(info);
*info = info_;
return info_;
}
AudioEncoder::EncodedInfo info_;
@ -144,7 +141,8 @@ TEST_F(AudioEncoderCopyRedTest, CheckImmediateEncode) {
InSequence s;
MockFunction<void(int check_point_id)> check;
for (int i = 1; i <= 6; ++i) {
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _));
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.WillRepeatedly(Return(AudioEncoder::EncodedInfo()));
EXPECT_CALL(check, Call(i));
Encode();
check.Call(i);
@ -153,13 +151,13 @@ TEST_F(AudioEncoderCopyRedTest, CheckImmediateEncode) {
// Checks that no output is produced if the underlying codec doesn't emit any
// new data, even if the RED codec is loaded with a secondary encoding.
TEST_F(AudioEncoderCopyRedTest, CheckNoOuput) {
TEST_F(AudioEncoderCopyRedTest, CheckNoOutput) {
// Start with one Encode() call that will produce output.
static const size_t kEncodedSize = 17;
AudioEncoder::EncodedInfo info;
info.encoded_bytes = kEncodedSize;
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.WillOnce(SetArgPointee<4>(info));
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.WillOnce(Return(info));
Encode();
// First call is a special case, since it does not include a secondary
// payload.
@ -168,15 +166,15 @@ TEST_F(AudioEncoderCopyRedTest, CheckNoOuput) {
// Next call to the speech encoder will not produce any output.
info.encoded_bytes = 0;
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.WillOnce(SetArgPointee<4>(info));
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.WillOnce(Return(info));
Encode();
EXPECT_EQ(0u, encoded_info_.encoded_bytes);
// Final call to the speech encoder will produce output.
info.encoded_bytes = kEncodedSize;
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.WillOnce(SetArgPointee<4>(info));
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.WillOnce(Return(info));
Encode();
EXPECT_EQ(2 * kEncodedSize, encoded_info_.encoded_bytes);
ASSERT_EQ(2u, encoded_info_.redundant.size());
@ -192,8 +190,8 @@ TEST_F(AudioEncoderCopyRedTest, CheckPayloadSizes) {
for (int encode_size = 1; encode_size <= kNumPackets; ++encode_size) {
AudioEncoder::EncodedInfo info;
info.encoded_bytes = encode_size;
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
.WillOnce(SetArgPointee<4>(info));
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.WillOnce(Return(info));
}
// First call is a special case, since it does not include a secondary
@ -218,7 +216,7 @@ TEST_F(AudioEncoderCopyRedTest, CheckTimestamps) {
helper.info_.encoded_bytes = 17;
helper.info_.encoded_timestamp = timestamp_;
uint32_t primary_timestamp = timestamp_;
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.WillRepeatedly(Invoke(&helper, &MockEncodeHelper::Encode));
// First call is a special case, since it does not include a secondary
@ -249,7 +247,7 @@ TEST_F(AudioEncoderCopyRedTest, CheckPayloads) {
payload[i] = i;
}
helper.payload_ = payload;
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.WillRepeatedly(Invoke(&helper, &MockEncodeHelper::Encode));
// First call is a special case, since it does not include a secondary
@ -286,7 +284,7 @@ TEST_F(AudioEncoderCopyRedTest, CheckPayloadType) {
helper.info_.encoded_bytes = 17;
const int primary_payload_type = red_payload_type_ + 1;
helper.info_.payload_type = primary_payload_type;
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _, _))
EXPECT_CALL(mock_encoder_, EncodeInternal(_, _, _, _))
.WillRepeatedly(Invoke(&helper, &MockEncodeHelper::Encode));
// First call is a special case, since it does not include a secondary

View File

@ -46,8 +46,8 @@ class AcmGenericCodecTest : public ::testing::Test {
int expected_send_even_if_empty) {
uint8_t out[kPacketSizeSamples];
AudioEncoder::EncodedInfo encoded_info;
codec_->GetAudioEncoder()->Encode(timestamp_, kZeroData, kDataLengthSamples,
kPacketSizeSamples, out, &encoded_info);
encoded_info = codec_->GetAudioEncoder()->Encode(
timestamp_, kZeroData, kDataLengthSamples, kPacketSizeSamples, out);
timestamp_ += kDataLengthSamples;
EXPECT_TRUE(encoded_info.redundant.empty());
EXPECT_EQ(expected_out_length, encoded_info.encoded_bytes);

View File

@ -246,9 +246,9 @@ int32_t AudioCodingModuleImpl::Encode(const InputData& input_data) {
last_rtp_timestamp_ = rtp_timestamp;
first_frame_ = false;
audio_encoder->Encode(rtp_timestamp, input_data.audio,
input_data.length_per_channel, sizeof(stream), stream,
&encoded_info);
encoded_info = audio_encoder->Encode(rtp_timestamp, input_data.audio,
input_data.length_per_channel,
sizeof(stream), stream);
if (encoded_info.encoded_bytes == 0 && !encoded_info.send_even_if_empty) {
// Not enough data.
return 0;

View File

@ -150,9 +150,9 @@ class AudioDecoderTest : public ::testing::Test {
samples_per_10ms, channels_,
interleaved_input.get());
audio_encoder_->Encode(0, interleaved_input.get(),
audio_encoder_->SampleRateHz() / 100,
data_length_ * 2, output, &encoded_info_);
encoded_info_ = audio_encoder_->Encode(
0, interleaved_input.get(), audio_encoder_->SampleRateHz() / 100,
data_length_ * 2, output);
}
EXPECT_EQ(payload_type_, encoded_info_.payload_type);
return static_cast<int>(encoded_info_.encoded_bytes);