turaj@webrtc.org 2014-11-11 01:44:13 +00:00
parent 05feff013e
commit 1431e4dd1c
8 changed files with 149 additions and 334 deletions

View File

@ -356,13 +356,11 @@ config("isac_config") {
source_set("isac") {
sources = [
"codecs/isac/main/interface/audio_encoder_isac.h",
"codecs/isac/main/interface/isac.h",
"codecs/isac/main/source/arith_routines.c",
"codecs/isac/main/source/arith_routines.h",
"codecs/isac/main/source/arith_routines_hist.c",
"codecs/isac/main/source/arith_routines_logist.c",
"codecs/isac/main/source/audio_encoder_isac.cc",
"codecs/isac/main/source/bandwidth_estimator.c",
"codecs/isac/main/source/bandwidth_estimator.h",
"codecs/isac/main/source/codec.h",
@ -409,12 +407,6 @@ source_set("isac") {
"codecs/isac/main/source/transform.c",
]
if (is_clang) {
# Suppress warnings from Chrome's Clang plugins.
# See http://code.google.com/p/webrtc/issues/detail?id=163 for details.
configs -= [ "//build/config/clang:find_bad_constructs" ]
}
if (is_linux) {
libs = [ "m" ]
}

View File

@ -1,106 +0,0 @@
/*
* 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_ISAC_MAIN_INTERFACE_AUDIO_ENCODER_ISAC_H_
#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_INTERFACE_AUDIO_ENCODER_ISAC_H_
#include <vector>
#include "webrtc/base/thread_annotations.h"
#include "webrtc/modules/audio_coding/codecs/audio_encoder.h"
#include "webrtc/modules/audio_coding/codecs/isac/main/interface/isac.h"
#include "webrtc/modules/audio_coding/neteq/interface/audio_decoder.h"
#include "webrtc/system_wrappers/interface/scoped_ptr.h"
namespace webrtc {
class CriticalSectionWrapper;
class AudioEncoderDecoderIsac : public AudioEncoder, public AudioDecoder {
public:
// For constructing an encoder in instantaneous mode. Allowed combinations
// are
// - 16000 Hz, 30 ms, 10000-32000 bps
// - 16000 Hz, 60 ms, 10000-32000 bps
// - 32000 Hz, 30 ms, 10000-56000 bps
struct Config {
Config();
bool IsOk() const;
int sample_rate_hz;
int frame_size_ms;
int bit_rate; // Limit on the short-term average bit rate, in bits/second.
};
// For constructing an encoder in channel-adaptive mode. The sample rate must
// be 16000 Hz; the initial frame size can be 30 or 60 ms; and the initial bit
// rate can be 10000-56000 bps.
struct ConfigAdaptive {
ConfigAdaptive();
bool IsOk() const;
int sample_rate_hz;
int initial_frame_size_ms;
int initial_bit_rate;
bool enforce_frame_size; // Prevent adaptive changes to the frame size?
};
explicit AudioEncoderDecoderIsac(const Config& config);
explicit AudioEncoderDecoderIsac(const ConfigAdaptive& config);
virtual ~AudioEncoderDecoderIsac() OVERRIDE;
// AudioEncoder public methods.
virtual int sample_rate_hz() const OVERRIDE;
virtual int num_channels() const OVERRIDE;
virtual int Num10MsFramesInNextPacket() const OVERRIDE;
// AudioDecoder methods.
virtual int Decode(const uint8_t* encoded,
size_t encoded_len,
int16_t* decoded,
SpeechType* speech_type) OVERRIDE;
virtual int DecodeRedundant(const uint8_t* encoded,
size_t encoded_len,
int16_t* decoded,
SpeechType* speech_type) OVERRIDE;
virtual bool HasDecodePlc() const OVERRIDE;
virtual int DecodePlc(int num_frames, int16_t* decoded) OVERRIDE;
virtual int Init() OVERRIDE;
virtual int IncomingPacket(const uint8_t* payload,
size_t payload_len,
uint16_t rtp_sequence_number,
uint32_t rtp_timestamp,
uint32_t arrival_timestamp) OVERRIDE;
virtual int ErrorCode() OVERRIDE;
protected:
// AudioEncoder protected method.
virtual bool Encode(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
uint32_t* encoded_timestamp) OVERRIDE;
private:
// iSAC encoder/decoder state, guarded by a mutex to ensure that encode calls
// from one thread won't clash with decode calls from another thread.
const scoped_ptr<CriticalSectionWrapper> lock_;
ISACStruct* isac_state_ GUARDED_BY(lock_);
// Have we accepted input but not yet emitted it in a packet?
bool packet_in_progress_;
// Timestamp of the first input of the currently in-progress packet.
uint32_t packet_timestamp_;
DISALLOW_COPY_AND_ASSIGN(AudioEncoderDecoderIsac);
};
} // namespace webrtc
#endif // WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_INTERFACE_AUDIO_ENCODER_ISAC_H_

View File

@ -1,188 +0,0 @@
/*
* 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/isac/main/interface/audio_encoder_isac.h"
#include "webrtc/modules/audio_coding/codecs/isac/main/interface/isac.h"
#include "webrtc/system_wrappers/interface/critical_section_wrapper.h"
namespace webrtc {
namespace {
int DivExact(int a, int b) {
CHECK_EQ(a % b, 0);
return a / b;
}
} // namespace
AudioEncoderDecoderIsac::Config::Config()
: sample_rate_hz(16000), frame_size_ms(30), bit_rate(32000) {}
bool AudioEncoderDecoderIsac::Config::IsOk() const {
switch (sample_rate_hz) {
case 16000:
return (frame_size_ms == 30 || frame_size_ms == 60) &&
bit_rate >= 10000 && bit_rate <= 32000;
case 32000:
return frame_size_ms == 30 && bit_rate >= 10000 && bit_rate <= 56000;
default:
return false;
}
}
AudioEncoderDecoderIsac::ConfigAdaptive::ConfigAdaptive()
: sample_rate_hz(16000),
initial_frame_size_ms(30),
initial_bit_rate(32000),
enforce_frame_size(false) {}
bool AudioEncoderDecoderIsac::ConfigAdaptive::IsOk() const {
return sample_rate_hz == 16000 &&
(initial_frame_size_ms == 30 || initial_frame_size_ms == 60) &&
initial_bit_rate >= 10000 && initial_bit_rate <= 56000;
}
AudioEncoderDecoderIsac::AudioEncoderDecoderIsac(const Config& config)
: lock_(CriticalSectionWrapper::CreateCriticalSection()),
packet_in_progress_(false) {
CHECK(config.IsOk());
CHECK_EQ(0, WebRtcIsac_Create(&isac_state_));
CHECK_EQ(0, WebRtcIsac_EncoderInit(isac_state_, 1));
CHECK_EQ(0, WebRtcIsac_SetEncSampRate(isac_state_, config.sample_rate_hz));
CHECK_EQ(0, WebRtcIsac_Control(isac_state_, config.bit_rate,
config.frame_size_ms));
CHECK_EQ(0, WebRtcIsac_SetDecSampRate(isac_state_, config.sample_rate_hz));
}
AudioEncoderDecoderIsac::AudioEncoderDecoderIsac(const ConfigAdaptive& config)
: lock_(CriticalSectionWrapper::CreateCriticalSection()),
packet_in_progress_(false) {
CHECK(config.IsOk());
CHECK_EQ(0, WebRtcIsac_Create(&isac_state_));
CHECK_EQ(0, WebRtcIsac_EncoderInit(isac_state_, 0));
CHECK_EQ(0, WebRtcIsac_SetEncSampRate(isac_state_, config.sample_rate_hz));
CHECK_EQ(0, WebRtcIsac_ControlBwe(isac_state_, config.initial_bit_rate,
config.initial_frame_size_ms,
config.enforce_frame_size));
CHECK_EQ(0, WebRtcIsac_SetDecSampRate(isac_state_, config.sample_rate_hz));
}
AudioEncoderDecoderIsac::~AudioEncoderDecoderIsac() {
CHECK_EQ(0, WebRtcIsac_Free(isac_state_));
}
int AudioEncoderDecoderIsac::sample_rate_hz() const {
CriticalSectionScoped cs(lock_.get());
return WebRtcIsac_EncSampRate(isac_state_);
}
int AudioEncoderDecoderIsac::num_channels() const {
return 1;
}
int AudioEncoderDecoderIsac::Num10MsFramesInNextPacket() const {
CriticalSectionScoped cs(lock_.get());
int samples_in_next_packet = WebRtcIsac_GetNewFrameLen(isac_state_);
return DivExact(samples_in_next_packet, DivExact(sample_rate_hz(), 100));
}
bool AudioEncoderDecoderIsac::Encode(uint32_t timestamp,
const int16_t* audio,
size_t max_encoded_bytes,
uint8_t* encoded,
size_t* encoded_bytes,
uint32_t* encoded_timestamp) {
if (!packet_in_progress_) {
// Starting a new packet; remember the timestamp for later.
packet_in_progress_ = true;
packet_timestamp_ = timestamp;
}
int r;
{
CriticalSectionScoped cs(lock_.get());
r = WebRtcIsac_Encode(isac_state_, audio, encoded);
}
if (r < 0) {
// An error occurred; propagate it to the caller.
packet_in_progress_ = false;
return false;
}
// WebRtcIsac_Encode doesn't allow us to tell it the size of the output
// 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;
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.
packet_in_progress_ = false;
*encoded_timestamp = packet_timestamp_;
}
return true;
}
int AudioEncoderDecoderIsac::Decode(const uint8_t* encoded,
size_t encoded_len,
int16_t* decoded,
SpeechType* speech_type) {
CriticalSectionScoped cs(lock_.get());
int16_t temp_type = 1; // Default is speech.
int16_t ret =
WebRtcIsac_Decode(isac_state_, encoded, static_cast<int16_t>(encoded_len),
decoded, &temp_type);
*speech_type = ConvertSpeechType(temp_type);
return ret;
}
int AudioEncoderDecoderIsac::DecodeRedundant(const uint8_t* encoded,
size_t encoded_len,
int16_t* decoded,
SpeechType* speech_type) {
CriticalSectionScoped cs(lock_.get());
int16_t temp_type = 1; // Default is speech.
int16_t ret = WebRtcIsac_DecodeRcu(isac_state_, encoded,
static_cast<int16_t>(encoded_len), decoded,
&temp_type);
*speech_type = ConvertSpeechType(temp_type);
return ret;
}
bool AudioEncoderDecoderIsac::HasDecodePlc() const { return true; }
int AudioEncoderDecoderIsac::DecodePlc(int num_frames, int16_t* decoded) {
CriticalSectionScoped cs(lock_.get());
return WebRtcIsac_DecodePlc(isac_state_, decoded, num_frames);
}
int AudioEncoderDecoderIsac::Init() {
CriticalSectionScoped cs(lock_.get());
return WebRtcIsac_DecoderInit(isac_state_);
}
int AudioEncoderDecoderIsac::IncomingPacket(const uint8_t* payload,
size_t payload_len,
uint16_t rtp_sequence_number,
uint32_t rtp_timestamp,
uint32_t arrival_timestamp) {
CriticalSectionScoped cs(lock_.get());
return WebRtcIsac_UpdateBwEstimate(
isac_state_, payload, static_cast<int32_t>(payload_len),
rtp_sequence_number, rtp_timestamp, arrival_timestamp);
}
int AudioEncoderDecoderIsac::ErrorCode() {
CriticalSectionScoped cs(lock_.get());
return WebRtcIsac_GetErrorCode(isac_state_);
}
} // namespace webrtc

View File

@ -25,12 +25,10 @@
],
},
'sources': [
'../interface/audio_encoder_isac.h',
'../interface/isac.h',
'arith_routines.c',
'arith_routines_hist.c',
'arith_routines_logist.c',
'audio_encoder_isac.cc',
'bandwidth_estimator.c',
'crc.c',
'decode.c',

View File

@ -13,7 +13,6 @@
#include <assert.h>
#include "webrtc/base/checks.h"
#include "webrtc/modules/audio_coding/codecs/isac/main/interface/audio_encoder_isac.h"
#include "webrtc/modules/audio_coding/neteq/audio_decoder_impl.h"
namespace webrtc {
@ -201,17 +200,11 @@ AudioDecoder* AudioDecoder::CreateAudioDecoder(NetEqDecoder codec_type) {
case kDecoderISAC:
return new AudioDecoderIsacFix;
#elif defined(WEBRTC_CODEC_ISAC)
case kDecoderISAC: {
AudioEncoderDecoderIsac::Config config;
config.sample_rate_hz = 16000;
return new AudioEncoderDecoderIsac(config);
}
case kDecoderISAC:
return new AudioDecoderIsac(16000);
case kDecoderISACswb:
case kDecoderISACfb: {
AudioEncoderDecoderIsac::Config config;
config.sample_rate_hz = 32000;
return new AudioEncoderDecoderIsac(config);
}
case kDecoderISACfb:
return new AudioDecoderIsac(32000);
#endif
#ifdef WEBRTC_CODEC_PCM16
case kDecoderPCM16B:

View File

@ -130,6 +130,67 @@ int AudioDecoderIlbc::Init() {
}
#endif
// iSAC float
#ifdef WEBRTC_CODEC_ISAC
AudioDecoderIsac::AudioDecoderIsac(int decode_sample_rate_hz) {
DCHECK(decode_sample_rate_hz == 16000 || decode_sample_rate_hz == 32000);
WebRtcIsac_Create(&isac_state_);
WebRtcIsac_SetDecSampRate(isac_state_, decode_sample_rate_hz);
}
AudioDecoderIsac::~AudioDecoderIsac() {
WebRtcIsac_Free(isac_state_);
}
int AudioDecoderIsac::Decode(const uint8_t* encoded, size_t encoded_len,
int16_t* decoded, SpeechType* speech_type) {
int16_t temp_type = 1; // Default is speech.
int16_t ret = WebRtcIsac_Decode(isac_state_,
encoded,
static_cast<int16_t>(encoded_len), decoded,
&temp_type);
*speech_type = ConvertSpeechType(temp_type);
return ret;
}
int AudioDecoderIsac::DecodeRedundant(const uint8_t* encoded,
size_t encoded_len, int16_t* decoded,
SpeechType* speech_type) {
int16_t temp_type = 1; // Default is speech.
int16_t ret = WebRtcIsac_DecodeRcu(isac_state_,
encoded,
static_cast<int16_t>(encoded_len), decoded,
&temp_type);
*speech_type = ConvertSpeechType(temp_type);
return ret;
}
int AudioDecoderIsac::DecodePlc(int num_frames, int16_t* decoded) {
return WebRtcIsac_DecodePlc(isac_state_, decoded, num_frames);
}
int AudioDecoderIsac::Init() {
return WebRtcIsac_DecoderInit(isac_state_);
}
int AudioDecoderIsac::IncomingPacket(const uint8_t* payload,
size_t payload_len,
uint16_t rtp_sequence_number,
uint32_t rtp_timestamp,
uint32_t arrival_timestamp) {
return WebRtcIsac_UpdateBwEstimate(isac_state_,
payload,
static_cast<int32_t>(payload_len),
rtp_sequence_number,
rtp_timestamp,
arrival_timestamp);
}
int AudioDecoderIsac::ErrorCode() {
return WebRtcIsac_GetErrorCode(isac_state_);
}
#endif
// iSAC fix
#ifdef WEBRTC_CODEC_ISACFX
AudioDecoderIsacFix::AudioDecoderIsacFix() {

View File

@ -130,6 +130,31 @@ class AudioDecoderIlbc : public AudioDecoder {
};
#endif
#ifdef WEBRTC_CODEC_ISAC
class AudioDecoderIsac : public AudioDecoder {
public:
explicit AudioDecoderIsac(int decode_sample_rate_hz);
virtual ~AudioDecoderIsac();
virtual int Decode(const uint8_t* encoded, size_t encoded_len,
int16_t* decoded, SpeechType* speech_type);
virtual int DecodeRedundant(const uint8_t* encoded, size_t encoded_len,
int16_t* decoded, SpeechType* speech_type);
virtual bool HasDecodePlc() const { return true; }
virtual int DecodePlc(int num_frames, int16_t* decoded);
virtual int Init();
virtual int IncomingPacket(const uint8_t* payload,
size_t payload_len,
uint16_t rtp_sequence_number,
uint32_t rtp_timestamp,
uint32_t arrival_timestamp);
virtual int ErrorCode();
private:
ISACStruct* isac_state_;
DISALLOW_COPY_AND_ASSIGN(AudioDecoderIsac);
};
#endif
#ifdef WEBRTC_CODEC_ISACFX
class AudioDecoderIsacFix : public AudioDecoder {
public:

View File

@ -25,7 +25,7 @@
#include "webrtc/modules/audio_coding/codecs/g722/include/g722_interface.h"
#include "webrtc/modules/audio_coding/codecs/ilbc/interface/ilbc.h"
#include "webrtc/modules/audio_coding/codecs/isac/fix/interface/isacfix.h"
#include "webrtc/modules/audio_coding/codecs/isac/main/interface/audio_encoder_isac.h"
#include "webrtc/modules/audio_coding/codecs/isac/main/interface/isac.h"
#include "webrtc/modules/audio_coding/codecs/opus/interface/audio_encoder_opus.h"
#include "webrtc/modules/audio_coding/codecs/pcm16b/include/pcm16b.h"
#include "webrtc/modules/audio_coding/neteq/tools/resample_input_audio_file.h"
@ -366,36 +366,76 @@ class AudioDecoderIsacFloatTest : public AudioDecoderTest {
protected:
AudioDecoderIsacFloatTest() : AudioDecoderTest() {
codec_input_rate_hz_ = 16000;
input_size_ = 160;
frame_size_ = 480;
data_length_ = 10 * frame_size_;
AudioEncoderDecoderIsac::Config config;
config.sample_rate_hz = codec_input_rate_hz_;
config.frame_size_ms =
1000 * static_cast<int>(frame_size_) / codec_input_rate_hz_;
// We need to create separate AudioEncoderDecoderIsac objects for encoding
// and decoding, because the test class destructor destroys them both.
audio_encoder_.reset(new AudioEncoderDecoderIsac(config));
decoder_ = new AudioEncoderDecoderIsac(config);
decoder_ = new AudioDecoderIsac(16000);
assert(decoder_);
WebRtcIsac_Create(&encoder_);
WebRtcIsac_SetEncSampRate(encoder_, 16000);
}
~AudioDecoderIsacFloatTest() {
WebRtcIsac_Free(encoder_);
}
virtual void InitEncoder() {
ASSERT_EQ(0, WebRtcIsac_EncoderInit(encoder_, 1)); // Fixed mode.
ASSERT_EQ(0, WebRtcIsac_Control(encoder_, 32000, 30)); // 32 kbps, 30 ms.
}
virtual int EncodeFrame(const int16_t* input, size_t input_len_samples,
uint8_t* output) {
// Insert 3 * 10 ms. Expect non-zero output on third call.
EXPECT_EQ(0, WebRtcIsac_Encode(encoder_, input, output));
input += input_size_;
EXPECT_EQ(0, WebRtcIsac_Encode(encoder_, input, output));
input += input_size_;
int enc_len_bytes = WebRtcIsac_Encode(encoder_, input, output);
EXPECT_GT(enc_len_bytes, 0);
return enc_len_bytes;
}
ISACStruct* encoder_;
int input_size_;
};
class AudioDecoderIsacSwbTest : public AudioDecoderTest {
protected:
AudioDecoderIsacSwbTest() : AudioDecoderTest() {
codec_input_rate_hz_ = 32000;
input_size_ = 320;
frame_size_ = 960;
data_length_ = 10 * frame_size_;
AudioEncoderDecoderIsac::Config config;
config.sample_rate_hz = codec_input_rate_hz_;
config.frame_size_ms =
1000 * static_cast<int>(frame_size_) / codec_input_rate_hz_;
// We need to create separate AudioEncoderDecoderIsac objects for encoding
// and decoding, because the test class destructor destroys them both.
audio_encoder_.reset(new AudioEncoderDecoderIsac(config));
decoder_ = new AudioEncoderDecoderIsac(config);
decoder_ = new AudioDecoderIsac(32000);
assert(decoder_);
WebRtcIsac_Create(&encoder_);
WebRtcIsac_SetEncSampRate(encoder_, 32000);
}
~AudioDecoderIsacSwbTest() {
WebRtcIsac_Free(encoder_);
}
virtual void InitEncoder() {
ASSERT_EQ(0, WebRtcIsac_EncoderInit(encoder_, 1)); // Fixed mode.
ASSERT_EQ(0, WebRtcIsac_Control(encoder_, 32000, 30)); // 32 kbps, 30 ms.
}
virtual int EncodeFrame(const int16_t* input, size_t input_len_samples,
uint8_t* output) {
// Insert 3 * 10 ms. Expect non-zero output on third call.
EXPECT_EQ(0, WebRtcIsac_Encode(encoder_, input, output));
input += input_size_;
EXPECT_EQ(0, WebRtcIsac_Encode(encoder_, input, output));
input += input_size_;
int enc_len_bytes = WebRtcIsac_Encode(encoder_, input, output);
EXPECT_GT(enc_len_bytes, 0);
return enc_len_bytes;
}
ISACStruct* encoder_;
int input_size_;
};
class AudioDecoderIsacFixTest : public AudioDecoderTest {