zqz/platform-external-webrtc
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Reland "Upconvert various types to int.", neteq portion.

Browse Source 
This reverts portions of commit cb180976dd0e9672cde4523d87b5f4857478b5e9, which
reverted commit 83ad33a8aed1fb00e422b6abd33c3e8942821c24.  Specifically, the
files in webrtc/modules/audio_coding/neteq/ are relanded.

The original commit message is below:

Upconvert various types to int.

Per comments from HL/kwiberg on https://webrtc-codereview.appspot.com/42569004 , when there is existing usage of mixed types (int16_t, int, etc.), we'd prefer to standardize on larger types like int and phase out use of int16_t.

Specifically, "Using int16 just because we're sure all reasonable values will fit in 16 bits isn't usually meaningful in C."

This converts some existing uses of int16_t (and, in a few cases, other types such as uint16_t) to int (or, in a few places, int32_t).  Other locations will be converted to size_t in a separate change.

BUG=none
TBR=kwiberg

Review URL: https://codereview.webrtc.org/1181073002

Cr-Commit-Position: refs/heads/master@{#9427}
This commit is contained in:
Peter Kasting
2015-06-11 19:57:18 -07:00
parent bc440d5651
commit 36b7cc3264
9 changed files with 44 additions and 46 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
webrtc/modules/audio_coding/neteq/audio_decoder_impl.cc
View File

@ -163,9 +163,9 @@ int AudioDecoderIlbc::DecodeInternal(const uint8_t* encoded,
SpeechType* speech_type) { SpeechType* speech_type) {
DCHECK_EQ(sample_rate_hz, 8000); DCHECK_EQ(sample_rate_hz, 8000);
int16_t temp_type = 1; // Default is speech. int16_t temp_type = 1; // Default is speech.
int16_t ret = WebRtcIlbcfix_Decode(dec_state_, encoded, int ret = WebRtcIlbcfix_Decode(dec_state_, encoded,
static_cast<int16_t>(encoded_len), decoded, static_cast<int16_t>(encoded_len), decoded,
&temp_type); &temp_type);
*speech_type = ConvertSpeechType(temp_type); *speech_type = ConvertSpeechType(temp_type);
return ret; return ret;
} }
@ -330,11 +330,11 @@ int AudioDecoderOpus::DecodeInternal(const uint8_t* encoded,
SpeechType* speech_type) { SpeechType* speech_type) {
DCHECK_EQ(sample_rate_hz, 48000); DCHECK_EQ(sample_rate_hz, 48000);
int16_t temp_type = 1; // Default is speech. int16_t temp_type = 1; // Default is speech.
int16_t ret = WebRtcOpus_Decode(dec_state_, encoded, int ret = WebRtcOpus_Decode(dec_state_, encoded,
static_cast<int16_t>(encoded_len), decoded, static_cast<int16_t>(encoded_len), decoded,
&temp_type); &temp_type);
if (ret > 0) if (ret > 0)
ret *= static_cast<int16_t>(channels_); // Return total number of samples. ret *= static_cast<int>(channels_); // Return total number of samples.
*speech_type = ConvertSpeechType(temp_type); *speech_type = ConvertSpeechType(temp_type);
return ret; return ret;
} }
@ -352,11 +352,11 @@ int AudioDecoderOpus::DecodeRedundantInternal(const uint8_t* encoded,
DCHECK_EQ(sample_rate_hz, 48000); DCHECK_EQ(sample_rate_hz, 48000);
int16_t temp_type = 1; // Default is speech. int16_t temp_type = 1; // Default is speech.
int16_t ret = WebRtcOpus_DecodeFec(dec_state_, encoded, int ret = WebRtcOpus_DecodeFec(dec_state_, encoded,
static_cast<int16_t>(encoded_len), decoded, static_cast<int16_t>(encoded_len), decoded,
&temp_type); &temp_type);
if (ret > 0) if (ret > 0)
ret *= static_cast<int16_t>(channels_); // Return total number of samples. ret *= static_cast<int>(channels_); // Return total number of samples.
*speech_type = ConvertSpeechType(temp_type); *speech_type = ConvertSpeechType(temp_type);
return ret; return ret;
} }

4
webrtc/modules/audio_coding/neteq/dsp_helper.cc
View File

@ -272,7 +272,7 @@ void DspHelper::CrossFade(const int16_t* input1, const int16_t* input2,
} }
void DspHelper::UnmuteSignal(const int16_t* input, size_t length, void DspHelper::UnmuteSignal(const int16_t* input, size_t length,
int16_t* factor, int16_t increment, int16_t* factor, int increment,
int16_t* output) { int16_t* output) {
uint16_t factor_16b = *factor; uint16_t factor_16b = *factor;
int32_t factor_32b = (static_cast<int32_t>(factor_16b) << 6) + 32; int32_t factor_32b = (static_cast<int32_t>(factor_16b) << 6) + 32;
@ -284,7 +284,7 @@ void DspHelper::UnmuteSignal(const int16_t* input, size_t length,
*factor = factor_16b; *factor = factor_16b;
} }
void DspHelper::MuteSignal(int16_t* signal, int16_t mute_slope, size_t length) { void DspHelper::MuteSignal(int16_t* signal, int mute_slope, size_t length) {
int32_t factor = (16384 << 6) + 32; int32_t factor = (16384 << 6) + 32;
for (size_t i = 0; i < length; i++) { for (size_t i = 0; i < length; i++) {
signal[i] = ((factor >> 6) * signal[i] + 8192) >> 14; signal[i] = ((factor >> 6) * signal[i] + 8192) >> 14;

4
webrtc/modules/audio_coding/neteq/dsp_helper.h
View File

@ -110,11 +110,11 @@ class DspHelper {
// sample and increases the gain by |increment| (Q20) for each sample. The // sample and increases the gain by |increment| (Q20) for each sample. The
// result is written to |output|. |length| samples are processed. // result is written to |output|. |length| samples are processed.
static void UnmuteSignal(const int16_t* input, size_t length, int16_t* factor, static void UnmuteSignal(const int16_t* input, size_t length, int16_t* factor,
int16_t increment, int16_t* output); int increment, int16_t* output);
// Starts at unity gain and gradually fades out |signal|. For each sample, // Starts at unity gain and gradually fades out |signal|. For each sample,
// the gain is reduced by |mute_slope| (Q14). |length| samples are processed. // the gain is reduced by |mute_slope| (Q14). |length| samples are processed.
static void MuteSignal(int16_t* signal, int16_t mute_slope, size_t length); static void MuteSignal(int16_t* signal, int mute_slope, size_t length);
// Downsamples |input| from |sample_rate_hz| to 4 kHz sample rate. The input // Downsamples |input| from |sample_rate_hz| to 4 kHz sample rate. The input
// has |input_length| samples, and the method will write |output_length| // has |input_length| samples, and the method will write |output_length|

42
webrtc/modules/audio_coding/neteq/expand.cc
View File

@ -239,14 +239,12 @@ int Expand::Process(AudioMultiVector* output) {
if (consecutive_expands_ == 3) { if (consecutive_expands_ == 3) {
// Let the mute factor decrease from 1.0 to 0.95 in 6.25 ms. // Let the mute factor decrease from 1.0 to 0.95 in 6.25 ms.
// mute_slope = 0.0010 / fs_mult in Q20. // mute_slope = 0.0010 / fs_mult in Q20.
parameters.mute_slope = std::max(parameters.mute_slope, parameters.mute_slope = std::max(parameters.mute_slope, 1049 / fs_mult);
static_cast<int16_t>(1049 / fs_mult));
} }
if (consecutive_expands_ == 7) { if (consecutive_expands_ == 7) {
// Let the mute factor decrease from 1.0 to 0.90 in 6.25 ms. // Let the mute factor decrease from 1.0 to 0.90 in 6.25 ms.
// mute_slope = 0.0020 / fs_mult in Q20. // mute_slope = 0.0020 / fs_mult in Q20.
parameters.mute_slope = std::max(parameters.mute_slope, parameters.mute_slope = std::max(parameters.mute_slope, 2097 / fs_mult);
static_cast<int16_t>(2097 / fs_mult));
} }
// Mute segment according to slope value. // Mute segment according to slope value.
@ -368,7 +366,7 @@ void Expand::AnalyzeSignal(int16_t* random_vector) {
InitializeForAnExpandPeriod(); InitializeForAnExpandPeriod();
// Calculate correlation in downsampled domain (4 kHz sample rate). // Calculate correlation in downsampled domain (4 kHz sample rate).
int16_t correlation_scale; int correlation_scale;
int correlation_length = 51; // TODO(hlundin): Legacy bit-exactness. int correlation_length = 51; // TODO(hlundin): Legacy bit-exactness.
// If it is decided to break bit-exactness |correlation_length| should be // If it is decided to break bit-exactness |correlation_length| should be
// initialized to the return value of Correlation(). // initialized to the return value of Correlation().
@ -445,7 +443,7 @@ void Expand::AnalyzeSignal(int16_t* random_vector) {
correlation_length + start_index + correlation_lags - 1); correlation_length + start_index + correlation_lags - 1);
correlation_scale = ((31 - WebRtcSpl_NormW32(signal_max * signal_max)) correlation_scale = ((31 - WebRtcSpl_NormW32(signal_max * signal_max))
+ (31 - WebRtcSpl_NormW32(correlation_length))) - 31; + (31 - WebRtcSpl_NormW32(correlation_length))) - 31;
correlation_scale = std::max(static_cast<int16_t>(0), correlation_scale); correlation_scale = std::max(0, correlation_scale);
// Calculate the correlation, store in |correlation_vector2|. // Calculate the correlation, store in |correlation_vector2|.
WebRtcSpl_CrossCorrelation( WebRtcSpl_CrossCorrelation(
@ -472,7 +470,7 @@ void Expand::AnalyzeSignal(int16_t* random_vector) {
// Calculate the correlation coefficient between the two portions of the // Calculate the correlation coefficient between the two portions of the
// signal. // signal.
int16_t corr_coefficient; int32_t corr_coefficient;
if ((energy1 > 0) && (energy2 > 0)) { if ((energy1 > 0) && (energy2 > 0)) {
int energy1_scale = std::max(16 - WebRtcSpl_NormW32(energy1), 0); int energy1_scale = std::max(16 - WebRtcSpl_NormW32(energy1), 0);
int energy2_scale = std::max(16 - WebRtcSpl_NormW32(energy2), 0); int energy2_scale = std::max(16 - WebRtcSpl_NormW32(energy2), 0);
@ -481,17 +479,17 @@ void Expand::AnalyzeSignal(int16_t* random_vector) {
// If sum is odd, add 1 to make it even. // If sum is odd, add 1 to make it even.
energy1_scale += 1; energy1_scale += 1;
} }
int16_t scaled_energy1 = energy1 >> energy1_scale; int32_t scaled_energy1 = energy1 >> energy1_scale;
int16_t scaled_energy2 = energy2 >> energy2_scale; int32_t scaled_energy2 = energy2 >> energy2_scale;
int16_t sqrt_energy_product = WebRtcSpl_SqrtFloor( int16_t sqrt_energy_product = static_cast<int16_t>(
scaled_energy1 * scaled_energy2); WebRtcSpl_SqrtFloor(scaled_energy1 * scaled_energy2));
// Calculate max_correlation / sqrt(energy1 * energy2) in Q14. // Calculate max_correlation / sqrt(energy1 * energy2) in Q14.
int cc_shift = 14 - (energy1_scale + energy2_scale) / 2; int cc_shift = 14 - (energy1_scale + energy2_scale) / 2;
max_correlation = WEBRTC_SPL_SHIFT_W32(max_correlation, cc_shift); max_correlation = WEBRTC_SPL_SHIFT_W32(max_correlation, cc_shift);
corr_coefficient = WebRtcSpl_DivW32W16(max_correlation, corr_coefficient = WebRtcSpl_DivW32W16(max_correlation,
sqrt_energy_product); sqrt_energy_product);
corr_coefficient = std::min(static_cast<int16_t>(16384), // Cap at 1.0 in Q14.
corr_coefficient); // Cap at 1.0 in Q14. corr_coefficient = std::min(16384, corr_coefficient);
} else { } else {
corr_coefficient = 0; corr_coefficient = 0;
} }
@ -512,8 +510,8 @@ void Expand::AnalyzeSignal(int16_t* random_vector) {
if ((energy1 / 4 < energy2) && (energy1 > energy2 / 4)) { if ((energy1 / 4 < energy2) && (energy1 > energy2 / 4)) {
// Energy constraint fulfilled. Use both vectors and scale them // Energy constraint fulfilled. Use both vectors and scale them
// accordingly. // accordingly.
int16_t scaled_energy2 = std::max(16 - WebRtcSpl_NormW32(energy2), 0); int32_t scaled_energy2 = std::max(16 - WebRtcSpl_NormW32(energy2), 0);
int16_t scaled_energy1 = scaled_energy2 - 13; int32_t scaled_energy1 = scaled_energy2 - 13;
// Calculate scaled_energy1 / scaled_energy2 in Q13. // Calculate scaled_energy1 / scaled_energy2 in Q13.
int32_t energy_ratio = WebRtcSpl_DivW32W16( int32_t energy_ratio = WebRtcSpl_DivW32W16(
WEBRTC_SPL_SHIFT_W32(energy1, -scaled_energy1), WEBRTC_SPL_SHIFT_W32(energy1, -scaled_energy1),
@ -682,7 +680,8 @@ void Expand::AnalyzeSignal(int16_t* random_vector) {
// voice_mix_factor = 0; // voice_mix_factor = 0;
if (corr_coefficient > 7875) { if (corr_coefficient > 7875) {
int16_t x1, x2, x3; int16_t x1, x2, x3;
x1 = corr_coefficient; // |corr_coefficient| is in Q14. // |corr_coefficient| is in Q14.
x1 = static_cast<int16_t>(corr_coefficient);
x2 = (x1 * x1) >> 14; // Shift 14 to keep result in Q14. x2 = (x1 * x1) >> 14; // Shift 14 to keep result in Q14.
x3 = (x1 * x2) >> 14; x3 = (x1 * x2) >> 14;
static const int kCoefficients[4] = { -5179, 19931, -16422, 5776 }; static const int kCoefficients[4] = { -5179, 19931, -16422, 5776 };
@ -709,7 +708,7 @@ void Expand::AnalyzeSignal(int16_t* random_vector) {
// the division. // the division.
// Shift the denominator from Q13 to Q5 before the division. The result of // Shift the denominator from Q13 to Q5 before the division. The result of
// the division will then be in Q20. // the division will then be in Q20.
int16_t temp_ratio = WebRtcSpl_DivW32W16( int temp_ratio = WebRtcSpl_DivW32W16(
(slope - 8192) << 12, (slope - 8192) << 12,
static_cast<int16_t>((distortion_lag * slope) >> 8)); static_cast<int16_t>((distortion_lag * slope) >> 8));
if (slope > 14746) { if (slope > 14746) {
@ -730,8 +729,7 @@ void Expand::AnalyzeSignal(int16_t* random_vector) {
// Make sure the mute factor decreases from 1.0 to 0.9 in no more than // Make sure the mute factor decreases from 1.0 to 0.9 in no more than
// 6.25 ms. // 6.25 ms.
// mute_slope >= 0.005 / fs_mult in Q20. // mute_slope >= 0.005 / fs_mult in Q20.
parameters.mute_slope = std::max(static_cast<int16_t>(5243 / fs_mult), parameters.mute_slope = std::max(5243 / fs_mult, parameters.mute_slope);
parameters.mute_slope);
} else if (slope > 8028) { } else if (slope > 8028) {
parameters.mute_slope = 0; parameters.mute_slope = 0;
} }
@ -755,7 +753,7 @@ Expand::ChannelParameters::ChannelParameters()
void Expand::Correlation(const int16_t* input, void Expand::Correlation(const int16_t* input,
size_t input_length, size_t input_length,
int16_t* output, int16_t* output,
int16_t* output_scale) const { int* output_scale) const {
// Set parameters depending on sample rate. // Set parameters depending on sample rate.
const int16_t* filter_coefficients; const int16_t* filter_coefficients;
int16_t num_coefficients; int16_t num_coefficients;
@ -844,7 +842,7 @@ Expand* ExpandFactory::Create(BackgroundNoise* background_noise,
// TODO(turajs): This can be moved to BackgroundNoise class. // TODO(turajs): This can be moved to BackgroundNoise class.
void Expand::GenerateBackgroundNoise(int16_t* random_vector, void Expand::GenerateBackgroundNoise(int16_t* random_vector,
size_t channel, size_t channel,
int16_t mute_slope, int mute_slope,
bool too_many_expands, bool too_many_expands,
size_t num_noise_samples, size_t num_noise_samples,
int16_t* buffer) { int16_t* buffer) {
@ -887,7 +885,7 @@ void Expand::GenerateBackgroundNoise(int16_t* random_vector,
bgn_mute_factor > 0) { bgn_mute_factor > 0) {
// Fade BGN to zero. // Fade BGN to zero.
// Calculate muting slope, approximately -2^18 / fs_hz. // Calculate muting slope, approximately -2^18 / fs_hz.
int16_t mute_slope; int mute_slope;
if (fs_hz_ == 8000) { if (fs_hz_ == 8000) {
mute_slope = -32; mute_slope = -32;
} else if (fs_hz_ == 16000) { } else if (fs_hz_ == 16000) {

6
webrtc/modules/audio_coding/neteq/expand.h
View File

@ -72,7 +72,7 @@ class Expand {
void GenerateBackgroundNoise(int16_t* random_vector, void GenerateBackgroundNoise(int16_t* random_vector,
size_t channel, size_t channel,
int16_t mute_slope, int mute_slope,
bool too_many_expands, bool too_many_expands,
size_t num_noise_samples, size_t num_noise_samples,
int16_t* buffer); int16_t* buffer);
@ -113,7 +113,7 @@ class Expand {
AudioVector expand_vector0; AudioVector expand_vector0;
AudioVector expand_vector1; AudioVector expand_vector1;
bool onset; bool onset;
int16_t mute_slope; /* Q20 */ int mute_slope; /* Q20 */
}; };
// Calculate the auto-correlation of |input|, with length |input_length| // Calculate the auto-correlation of |input|, with length |input_length|
@ -123,7 +123,7 @@ class Expand {
void Correlation(const int16_t* input, void Correlation(const int16_t* input,
size_t input_length, size_t input_length,
int16_t* output, int16_t* output,
int16_t* output_scale) const; int* output_scale) const;
void UpdateLagIndex(); void UpdateLagIndex();

4
webrtc/modules/audio_coding/neteq/merge.cc
View File

@ -314,7 +314,7 @@ int16_t Merge::CorrelateAndPeakSearch(int16_t expanded_max, int16_t input_max,
const int max_corr_length = kMaxCorrelationLength; const int max_corr_length = kMaxCorrelationLength;
int stop_position_downsamp = int stop_position_downsamp =
std::min(max_corr_length, expand_->max_lag() / (fs_mult_ * 2) + 1); std::min(max_corr_length, expand_->max_lag() / (fs_mult_ * 2) + 1);
int16_t correlation_shift = 0; int correlation_shift = 0;
if (expanded_max * input_max > 26843546) { if (expanded_max * input_max > 26843546) {
correlation_shift = 3; correlation_shift = 3;
} }
@ -333,7 +333,7 @@ int16_t Merge::CorrelateAndPeakSearch(int16_t expanded_max, int16_t input_max,
int16_t* correlation_ptr = &correlation16[pad_length]; int16_t* correlation_ptr = &correlation16[pad_length];
int32_t max_correlation = WebRtcSpl_MaxAbsValueW32(correlation, int32_t max_correlation = WebRtcSpl_MaxAbsValueW32(correlation,
stop_position_downsamp); stop_position_downsamp);
int16_t norm_shift = std::max(0, 17 - WebRtcSpl_NormW32(max_correlation)); int norm_shift = std::max(0, 17 - WebRtcSpl_NormW32(max_correlation));
WebRtcSpl_VectorBitShiftW32ToW16(correlation_ptr, stop_position_downsamp, WebRtcSpl_VectorBitShiftW32ToW16(correlation_ptr, stop_position_downsamp,
correlation, norm_shift); correlation, norm_shift);

2
webrtc/modules/audio_coding/neteq/neteq_impl.cc
View File

@ -1278,7 +1278,7 @@ int NetEqImpl::DecodeLoop(PacketList* packet_list, Operations* operation,
*operation == kPreemptiveExpand); *operation == kPreemptiveExpand);
packet_list->pop_front(); packet_list->pop_front();
size_t payload_length = packet->payload_length; size_t payload_length = packet->payload_length;
int16_t decode_length; int decode_length;
if (packet->sync_packet) { if (packet->sync_packet) {
// Decode to silence with the same frame size as the last decode. // Decode to silence with the same frame size as the last decode.
LOG(LS_VERBOSE) << "Decoding sync-packet: " << LOG(LS_VERBOSE) << "Decoding sync-packet: " <<

4
webrtc/modules/audio_coding/neteq/normal.cc
View File

@ -111,7 +111,7 @@ int Normal::Process(const int16_t* input,
} }
// If muted increase by 0.64 for every 20 ms (NB/WB 0.0040/0.0020 in Q14). // If muted increase by 0.64 for every 20 ms (NB/WB 0.0040/0.0020 in Q14).
int16_t increment = 64 / fs_mult; int increment = static_cast<int>(64 / fs_mult);
for (size_t i = 0; i < length_per_channel; i++) { for (size_t i = 0; i < length_per_channel; i++) {
// Scale with mute factor. // Scale with mute factor.
assert(channel_ix < output->Channels()); assert(channel_ix < output->Channels());
@ -178,7 +178,7 @@ int Normal::Process(const int16_t* input,
// Previous was neither of Expand, FadeToBGN or RFC3389_CNG, but we are // Previous was neither of Expand, FadeToBGN or RFC3389_CNG, but we are
// still ramping up from previous muting. // still ramping up from previous muting.
// If muted increase by 0.64 for every 20 ms (NB/WB 0.0040/0.0020 in Q14). // If muted increase by 0.64 for every 20 ms (NB/WB 0.0040/0.0020 in Q14).
int16_t increment = 64 / fs_mult; int increment = static_cast<int>(64 / fs_mult);
size_t length_per_channel = length / output->Channels(); size_t length_per_channel = length / output->Channels();
for (size_t i = 0; i < length_per_channel; i++) { for (size_t i = 0; i < length_per_channel; i++) {
for (size_t channel_ix = 0; channel_ix < output->Channels(); for (size_t channel_ix = 0; channel_ix < output->Channels();

2
webrtc/modules/audio_coding/neteq/test/RTPencode.cc
View File

@ -1561,7 +1561,7 @@ int NetEQTest_encode(int coder,
int useVAD, int useVAD,
int bitrate, int bitrate,
int numChannels) { int numChannels) {
short cdlen = 0; int cdlen = 0;
int16_t* tempdata; int16_t* tempdata;
static int first_cng = 1; static int first_cng = 1;
int16_t tempLen; int16_t tempLen;