generic-library/webrtc
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Since NetEq4 is ready to handle 48 kHz codec, it is good to remove the 48-to-32kHz downsampling of Opus output. This facilitates webrtc to make full use of Opus's bandwidth and eliminates unneeded computation in resampling.

Browse Source 
TEST=passed_all_trybots
R=henrik.lundin@webrtc.org, tina.legrand@webrtc.org

Review URL: https://webrtc-codereview.appspot.com/16619005

git-svn-id: http://webrtc.googlecode.com/svn/trunk@6458 4adac7df-926f-26a2-2b94-8c16560cd09d
This commit is contained in:
minyue@webrtc.org
2014-06-17 08:02:05 +00:00
parent 44a317a698
commit 8f8503d947
12 changed files with 249 additions and 433 deletions
Download Patch File Download Diff File Expand all files Collapse all files

241
webrtc/modules/audio_coding/codecs/opus/opus_unittest.cc
View File

@@ -19,9 +19,13 @@ struct WebRtcOpusDecInst;
namespace webrtc {
// Number of samples in a 60 ms stereo frame, sampled at 48 kHz.
const int kOpusNumberOfSamples = 480 * 6 * 2;
const int kOpusMaxFrameSamples = 48 * 60 * 2;
// Maximum number of bytes in output bitstream.
const size_t kMaxBytes = 1000;
// Number of samples-per-channel in a 20 ms frame, sampled at 48 kHz.
const int kOpus20msFrameSamples = 48 * 20;
// Number of samples-per-channel in a 10 ms frame, sampled at 48 kHz.
const int kOpus10msFrameSamples = 48 * 10;
class OpusTest : public ::testing::Test {
protected:
@@ -35,8 +39,8 @@ class OpusTest : public ::testing::Test {
WebRtcOpusDecInst* opus_stereo_decoder_;
WebRtcOpusDecInst* opus_stereo_decoder_new_;
int16_t speech_data_[kOpusNumberOfSamples];
int16_t output_data_[kOpusNumberOfSamples];
int16_t speech_data_[kOpusMaxFrameSamples];
int16_t output_data_[kOpusMaxFrameSamples];
uint8_t bitstream_[kMaxBytes];
};
@@ -50,17 +54,14 @@ OpusTest::OpusTest()
}
void OpusTest::SetUp() {
// Read some samples from a speech file, to be used in the encode test.
// In this test we do not care that the sampling frequency of the file is
// really 32000 Hz. We pretend that it is 48000 Hz.
FILE* input_file;
const std::string file_name =
webrtc::test::ResourcePath("audio_coding/testfile32kHz", "pcm");
webrtc::test::ResourcePath("audio_coding/speech_mono_32_48kHz", "pcm");
input_file = fopen(file_name.c_str(), "rb");
ASSERT_TRUE(input_file != NULL);
ASSERT_EQ(kOpusNumberOfSamples,
ASSERT_EQ(kOpusMaxFrameSamples,
static_cast<int32_t>(fread(speech_data_, sizeof(int16_t),
kOpusNumberOfSamples, input_file)));
kOpusMaxFrameSamples, input_file)));
fclose(input_file);
input_file = NULL;
}
@@ -114,21 +115,24 @@ TEST_F(OpusTest, OpusEncodeDecodeMono) {
// Encode & decode.
int16_t encoded_bytes;
int16_t audio_type;
int16_t output_data_decode_new[kOpusNumberOfSamples];
int16_t output_data_decode[kOpusNumberOfSamples];
int16_t output_data_decode_new[kOpusMaxFrameSamples];
int16_t output_data_decode[kOpusMaxFrameSamples];
int16_t* coded = reinterpret_cast<int16_t*>(bitstream_);
encoded_bytes = WebRtcOpus_Encode(opus_mono_encoder_, speech_data_, 960,
kMaxBytes, bitstream_);
EXPECT_EQ(640, WebRtcOpus_DecodeNew(opus_mono_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(640, WebRtcOpus_Decode(opus_mono_decoder_, coded,
encoded_bytes, output_data_decode,
&audio_type));
encoded_bytes = WebRtcOpus_Encode(opus_mono_encoder_, speech_data_,
kOpus20msFrameSamples, kMaxBytes,
bitstream_);
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodeNew(opus_mono_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_Decode(opus_mono_decoder_, coded,
encoded_bytes, output_data_decode,
&audio_type));
// Data in |output_data_decode_new| should be the same as in
// |output_data_decode|.
for (int i = 0; i < 640; i++) {
for (int i = 0; i < kOpus20msFrameSamples; i++) {
EXPECT_EQ(output_data_decode_new[i], output_data_decode[i]);
}
@@ -154,26 +158,30 @@ TEST_F(OpusTest, OpusEncodeDecodeStereo) {
// Encode & decode.
int16_t encoded_bytes;
int16_t audio_type;
int16_t output_data_decode_new[kOpusNumberOfSamples];
int16_t output_data_decode[kOpusNumberOfSamples];
int16_t output_data_decode_slave[kOpusNumberOfSamples];
int16_t output_data_decode_new[kOpusMaxFrameSamples];
int16_t output_data_decode[kOpusMaxFrameSamples];
int16_t output_data_decode_slave[kOpusMaxFrameSamples];
int16_t* coded = reinterpret_cast<int16_t*>(bitstream_);
encoded_bytes = WebRtcOpus_Encode(opus_stereo_encoder_, speech_data_, 960,
kMaxBytes, bitstream_);
EXPECT_EQ(640, WebRtcOpus_DecodeNew(opus_stereo_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(640, WebRtcOpus_Decode(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode,
encoded_bytes = WebRtcOpus_Encode(opus_stereo_encoder_, speech_data_,
kOpus20msFrameSamples, kMaxBytes,
bitstream_);
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodeNew(opus_stereo_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_Decode(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode,
&audio_type));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodeSlave(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode_slave,
&audio_type));
EXPECT_EQ(640, WebRtcOpus_DecodeSlave(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode_slave,
&audio_type));
// Data in |output_data_decode_new| should be the same as in
// |output_data_decode| and |output_data_decode_slave| interleaved to a
// stereo signal.
for (int i = 0; i < 640; i++) {
for (int i = 0; i < kOpus20msFrameSamples; i++) {
EXPECT_EQ(output_data_decode_new[i * 2], output_data_decode[i]);
EXPECT_EQ(output_data_decode_new[i * 2 + 1], output_data_decode_slave[i]);
}
@@ -234,26 +242,30 @@ TEST_F(OpusTest, OpusDecodeInit) {
// Encode & decode.
int16_t encoded_bytes;
int16_t audio_type;
int16_t output_data_decode_new[kOpusNumberOfSamples];
int16_t output_data_decode[kOpusNumberOfSamples];
int16_t output_data_decode_slave[kOpusNumberOfSamples];
int16_t output_data_decode_new[kOpusMaxFrameSamples];
int16_t output_data_decode[kOpusMaxFrameSamples];
int16_t output_data_decode_slave[kOpusMaxFrameSamples];
int16_t* coded = reinterpret_cast<int16_t*>(bitstream_);
encoded_bytes = WebRtcOpus_Encode(opus_stereo_encoder_, speech_data_, 960,
kMaxBytes, bitstream_);
EXPECT_EQ(640, WebRtcOpus_DecodeNew(opus_stereo_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(640, WebRtcOpus_Decode(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode,
encoded_bytes = WebRtcOpus_Encode(opus_stereo_encoder_, speech_data_,
kOpus20msFrameSamples, kMaxBytes,
bitstream_);
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodeNew(opus_stereo_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_Decode(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode,
&audio_type));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodeSlave(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode_slave,
&audio_type));
EXPECT_EQ(640, WebRtcOpus_DecodeSlave(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode_slave,
&audio_type));
// Data in |output_data_decode_new| should be the same as in
// |output_data_decode| and |output_data_decode_slave| interleaved to a
// stereo signal.
for (int i = 0; i < 640; i++) {
for (int i = 0; i < kOpus20msFrameSamples; i++) {
EXPECT_EQ(output_data_decode_new[i * 2], output_data_decode[i]);
EXPECT_EQ(output_data_decode_new[i * 2 + 1], output_data_decode_slave[i]);
}
@@ -262,20 +274,23 @@ TEST_F(OpusTest, OpusDecodeInit) {
EXPECT_EQ(0, WebRtcOpus_DecoderInit(opus_stereo_decoder_));
EXPECT_EQ(0, WebRtcOpus_DecoderInitSlave(opus_stereo_decoder_));
EXPECT_EQ(640, WebRtcOpus_DecodeNew(opus_stereo_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(640, WebRtcOpus_Decode(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode,
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodeNew(opus_stereo_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_Decode(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode,
&audio_type));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodeSlave(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode_slave,
&audio_type));
EXPECT_EQ(640, WebRtcOpus_DecodeSlave(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode_slave,
&audio_type));
// Data in |output_data_decode_new| should be the same as in
// |output_data_decode| and |output_data_decode_slave| interleaved to a
// stereo signal.
for (int i = 0; i < 640; i++) {
for (int i = 0; i < kOpus20msFrameSamples; i++) {
EXPECT_EQ(output_data_decode_new[i * 2], output_data_decode[i]);
EXPECT_EQ(output_data_decode_new[i * 2 + 1], output_data_decode_slave[i]);
}
@@ -344,27 +359,31 @@ TEST_F(OpusTest, OpusDecodePlcMono) {
// Encode & decode.
int16_t encoded_bytes;
int16_t audio_type;
int16_t output_data_decode_new[kOpusNumberOfSamples];
int16_t output_data_decode[kOpusNumberOfSamples];
int16_t output_data_decode_new[kOpusMaxFrameSamples];
int16_t output_data_decode[kOpusMaxFrameSamples];
int16_t* coded = reinterpret_cast<int16_t*>(bitstream_);
encoded_bytes = WebRtcOpus_Encode(opus_mono_encoder_, speech_data_, 960,
kMaxBytes, bitstream_);
EXPECT_EQ(640, WebRtcOpus_DecodeNew(opus_mono_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(640, WebRtcOpus_Decode(opus_mono_decoder_, coded,
encoded_bytes, output_data_decode,
&audio_type));
encoded_bytes = WebRtcOpus_Encode(opus_mono_encoder_, speech_data_,
kOpus20msFrameSamples, kMaxBytes,
bitstream_);
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodeNew(opus_mono_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_Decode(opus_mono_decoder_, coded,
encoded_bytes, output_data_decode,
&audio_type));
// Call decoder PLC for both versions of the decoder.
int16_t plc_buffer[kOpusNumberOfSamples];
int16_t plc_buffer_new[kOpusNumberOfSamples];
EXPECT_EQ(640, WebRtcOpus_DecodePlcMaster(opus_mono_decoder_, plc_buffer, 1));
EXPECT_EQ(640, WebRtcOpus_DecodePlc(opus_mono_decoder_new_,
plc_buffer_new, 1));
int16_t plc_buffer[kOpusMaxFrameSamples];
int16_t plc_buffer_new[kOpusMaxFrameSamples];
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodePlcMaster(opus_mono_decoder_, plc_buffer, 1));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodePlc(opus_mono_decoder_new_, plc_buffer_new, 1));
// Data in |plc_buffer| should be the same as in |plc_buffer_new|.
for (int i = 0; i < 640; i++) {
for (int i = 0; i < kOpus20msFrameSamples; i++) {
EXPECT_EQ(plc_buffer[i], plc_buffer_new[i]);
}
@@ -391,36 +410,42 @@ TEST_F(OpusTest, OpusDecodePlcStereo) {
// Encode & decode.
int16_t encoded_bytes;
int16_t audio_type;
int16_t output_data_decode_new[kOpusNumberOfSamples];
int16_t output_data_decode[kOpusNumberOfSamples];
int16_t output_data_decode_slave[kOpusNumberOfSamples];
int16_t output_data_decode_new[kOpusMaxFrameSamples];
int16_t output_data_decode[kOpusMaxFrameSamples];
int16_t output_data_decode_slave[kOpusMaxFrameSamples];
int16_t* coded = reinterpret_cast<int16_t*>(bitstream_);
encoded_bytes = WebRtcOpus_Encode(opus_stereo_encoder_, speech_data_, 960,
kMaxBytes, bitstream_);
EXPECT_EQ(640, WebRtcOpus_DecodeNew(opus_stereo_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(640, WebRtcOpus_Decode(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode,
encoded_bytes = WebRtcOpus_Encode(opus_stereo_encoder_, speech_data_,
kOpus20msFrameSamples, kMaxBytes,
bitstream_);
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodeNew(opus_stereo_decoder_new_, bitstream_,
encoded_bytes, output_data_decode_new,
&audio_type));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_Decode(opus_stereo_decoder_, coded,
encoded_bytes, output_data_decode,
&audio_type));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodeSlave(opus_stereo_decoder_, coded,
encoded_bytes,
output_data_decode_slave,
&audio_type));
EXPECT_EQ(640, WebRtcOpus_DecodeSlave(opus_stereo_decoder_, coded,
encoded_bytes,
output_data_decode_slave,
&audio_type));
// Call decoder PLC for both versions of the decoder.
int16_t plc_buffer_left[kOpusNumberOfSamples];
int16_t plc_buffer_right[kOpusNumberOfSamples];
int16_t plc_buffer_new[kOpusNumberOfSamples];
EXPECT_EQ(640, WebRtcOpus_DecodePlcMaster(opus_stereo_decoder_,
plc_buffer_left, 1));
EXPECT_EQ(640, WebRtcOpus_DecodePlcSlave(opus_stereo_decoder_,
plc_buffer_right, 1));
EXPECT_EQ(640, WebRtcOpus_DecodePlc(opus_stereo_decoder_new_, plc_buffer_new,
1));
int16_t plc_buffer_left[kOpusMaxFrameSamples];
int16_t plc_buffer_right[kOpusMaxFrameSamples];
int16_t plc_buffer_new[kOpusMaxFrameSamples];
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodePlcMaster(opus_stereo_decoder_,
plc_buffer_left, 1));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodePlcSlave(opus_stereo_decoder_,
plc_buffer_right, 1));
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DecodePlc(opus_stereo_decoder_new_, plc_buffer_new, 1));
// Data in |plc_buffer_left| and |plc_buffer_right|should be the same as the
// interleaved samples in |plc_buffer_new|.
for (int i = 0, j = 0; i < 640; i++) {
for (int i = 0, j = 0; i < kOpus20msFrameSamples; i++) {
EXPECT_EQ(plc_buffer_left[i], plc_buffer_new[j++]);
EXPECT_EQ(plc_buffer_right[i], plc_buffer_new[j++]);
}
@@ -437,21 +462,23 @@ TEST_F(OpusTest, OpusDurationEstimation) {
EXPECT_EQ(0, WebRtcOpus_EncoderCreate(&opus_stereo_encoder_, 2));
EXPECT_EQ(0, WebRtcOpus_DecoderCreate(&opus_stereo_decoder_, 2));
// Encode with different packet sizes (input 48 kHz, output in 32 kHz).
int16_t encoded_bytes;
// 10 ms.
encoded_bytes = WebRtcOpus_Encode(opus_stereo_encoder_, speech_data_, 480,
kMaxBytes, bitstream_);
EXPECT_EQ(320, WebRtcOpus_DurationEst(opus_stereo_decoder_, bitstream_,
encoded_bytes));
encoded_bytes = WebRtcOpus_Encode(opus_stereo_encoder_, speech_data_,
kOpus10msFrameSamples, kMaxBytes,
bitstream_);
EXPECT_EQ(kOpus10msFrameSamples,
WebRtcOpus_DurationEst(opus_stereo_decoder_, bitstream_,
encoded_bytes));
// 20 ms
encoded_bytes = WebRtcOpus_Encode(opus_stereo_encoder_, speech_data_, 960,
kMaxBytes, bitstream_);
EXPECT_EQ(640, WebRtcOpus_DurationEst(opus_stereo_decoder_, bitstream_,
encoded_bytes));
encoded_bytes = WebRtcOpus_Encode(opus_stereo_encoder_, speech_data_,
kOpus20msFrameSamples, kMaxBytes,
bitstream_);
EXPECT_EQ(kOpus20msFrameSamples,
WebRtcOpus_DurationEst(opus_stereo_decoder_, bitstream_,
encoded_bytes));
// Free memory.
EXPECT_EQ(0, WebRtcOpus_EncoderFree(opus_stereo_encoder_));