diff --git a/webrtc/modules/audio_coding/neteq4/decision_logic.cc b/webrtc/modules/audio_coding/neteq4/decision_logic.cc index 58accfd1b..04b886a2e 100644 --- a/webrtc/modules/audio_coding/neteq4/decision_logic.cc +++ b/webrtc/modules/audio_coding/neteq4/decision_logic.cc @@ -128,7 +128,7 @@ Operations DecisionLogic::GetDecision(const SyncBuffer& sync_buffer, const int cur_size_samples = samples_left + packet_buffer_.NumSamplesInBuffer(decoder_database_, decoder_frame_length); - NETEQ_LOG_VERBOSE << "Buffers: " << packet_buffer_.NumPacketsInBuffer() << + LOG(LS_VERBOSE) << "Buffers: " << packet_buffer_.NumPacketsInBuffer() << " packets * " << decoder_frame_length << " samples/packet + " << samples_left << " samples in sync buffer = " << cur_size_samples; diff --git a/webrtc/modules/audio_coding/neteq4/defines.h b/webrtc/modules/audio_coding/neteq4/defines.h index 67b7cde3e..b6f9eb2bc 100644 --- a/webrtc/modules/audio_coding/neteq4/defines.h +++ b/webrtc/modules/audio_coding/neteq4/defines.h @@ -47,11 +47,5 @@ enum Modes { kModeUndefined = -1 }; -#ifdef NETEQ4_VERBOSE_LOGGING -#define NETEQ_LOG_VERBOSE LOG(LS_VERBOSE) -#else -#define NETEQ_LOG_VERBOSE while(false)LOG(LS_VERBOSE) -#endif - } // namespace webrtc #endif // WEBRTC_MODULES_AUDIO_CODING_NETEQ4_DEFINES_H_ diff --git a/webrtc/modules/audio_coding/neteq4/neteq_impl.cc b/webrtc/modules/audio_coding/neteq4/neteq_impl.cc index d872b80eb..8e8ffe24c 100644 --- a/webrtc/modules/audio_coding/neteq4/neteq_impl.cc +++ b/webrtc/modules/audio_coding/neteq4/neteq_impl.cc @@ -124,7 +124,7 @@ int NetEqImpl::InsertPacket(const WebRtcRTPHeader& rtp_header, int length_bytes, uint32_t receive_timestamp) { CriticalSectionScoped lock(crit_sect_.get()); - NETEQ_LOG_VERBOSE << "InsertPacket: ts=" << rtp_header.header.timestamp << + LOG(LS_VERBOSE) << "InsertPacket: ts=" << rtp_header.header.timestamp << ", sn=" << rtp_header.header.sequenceNumber << ", pt=" << static_cast(rtp_header.header.payloadType) << ", ssrc=" << rtp_header.header.ssrc << @@ -143,10 +143,10 @@ int NetEqImpl::GetAudio(size_t max_length, int16_t* output_audio, int* samples_per_channel, int* num_channels, NetEqOutputType* type) { CriticalSectionScoped lock(crit_sect_.get()); - NETEQ_LOG_VERBOSE << "GetAudio"; + LOG(LS_VERBOSE) << "GetAudio"; int error = GetAudioInternal(max_length, output_audio, samples_per_channel, num_channels); - NETEQ_LOG_VERBOSE << "Produced " << *samples_per_channel << + LOG(LS_VERBOSE) << "Produced " << *samples_per_channel << " samples/channel for " << *num_channels << " channel(s)"; if (error != 0) { LOG_FERR1(LS_WARNING, GetAudioInternal, error); @@ -623,7 +623,7 @@ int NetEqImpl::GetAudioInternal(size_t max_length, int16_t* output, last_mode_ = kModeError; return return_value; } - NETEQ_LOG_VERBOSE << "GetDecision returned operation=" << operation << + LOG(LS_VERBOSE) << "GetDecision returned operation=" << operation << " and " << packet_list.size() << " packet(s)"; AudioDecoder::SpeechType speech_type; @@ -735,7 +735,7 @@ int NetEqImpl::GetAudioInternal(size_t max_length, int16_t* output, sync_buffer_->GetNextAudioInterleaved(num_output_samples_per_channel, output)); *num_channels = static_cast(sync_buffer_->Channels()); - NETEQ_LOG_VERBOSE << "Sync buffer (" << *num_channels << " channel(s)):" << + LOG(LS_VERBOSE) << "Sync buffer (" << *num_channels << " channel(s)):" << " insert " << algorithm_buffer_->Size() << " samples, extract " << samples_from_sync << " samples"; if (samples_from_sync != output_size_samples_) { @@ -1162,7 +1162,7 @@ int NetEqImpl::DecodeLoop(PacketList* packet_list, Operations* operation, int16_t decode_length; if (!packet->primary) { // This is a redundant payload; call the special decoder method. - NETEQ_LOG_VERBOSE << "Decoding packet (redundant):" << + LOG(LS_VERBOSE) << "Decoding packet (redundant):" << " ts=" << packet->header.timestamp << ", sn=" << packet->header.sequenceNumber << ", pt=" << static_cast(packet->header.payloadType) << @@ -1172,7 +1172,7 @@ int NetEqImpl::DecodeLoop(PacketList* packet_list, Operations* operation, packet->payload, packet->payload_length, &decoded_buffer_[*decoded_length], speech_type); } else { - NETEQ_LOG_VERBOSE << "Decoding packet: ts=" << packet->header.timestamp << + LOG(LS_VERBOSE) << "Decoding packet: ts=" << packet->header.timestamp << ", sn=" << packet->header.sequenceNumber << ", pt=" << static_cast(packet->header.payloadType) << ", ssrc=" << packet->header.ssrc << @@ -1190,7 +1190,7 @@ int NetEqImpl::DecodeLoop(PacketList* packet_list, Operations* operation, // Update |decoder_frame_length_| with number of samples per channel. decoder_frame_length_ = decode_length / static_cast(decoder->channels()); - NETEQ_LOG_VERBOSE << "Decoded " << decode_length << " samples (" << + LOG(LS_VERBOSE) << "Decoded " << decode_length << " samples (" << decoder->channels() << " channel(s) -> " << decoder_frame_length_ << " samples per channel)"; } else if (decode_length < 0) { diff --git a/webrtc/modules/audio_coding/neteq4/timestamp_scaler.cc b/webrtc/modules/audio_coding/neteq4/timestamp_scaler.cc index 423edeeb4..b2b5b40a3 100644 --- a/webrtc/modules/audio_coding/neteq4/timestamp_scaler.cc +++ b/webrtc/modules/audio_coding/neteq4/timestamp_scaler.cc @@ -85,7 +85,7 @@ uint32_t TimestampScaler::ToInternal(uint32_t external_timestamp, assert(denominator_ > 0); // Should not be possible. external_ref_ = external_timestamp; internal_ref_ += (external_diff * numerator_) / denominator_; - NETEQ_LOG_VERBOSE << "Converting timestamp: " << external_timestamp << + LOG(LS_VERBOSE) << "Converting timestamp: " << external_timestamp << " -> " << internal_ref_; return internal_ref_; } else {