| /* |
| * Copyright (c) 2013 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_MAIN_ACM2_ACM_RECEIVER_H_ |
| #define WEBRTC_MODULES_AUDIO_CODING_MAIN_ACM2_ACM_RECEIVER_H_ |
| |
| #include <map> |
| #include <vector> |
| |
| #include "webrtc/base/scoped_ptr.h" |
| #include "webrtc/base/thread_annotations.h" |
| #include "webrtc/common_audio/vad/include/webrtc_vad.h" |
| #include "webrtc/engine_configurations.h" |
| #include "webrtc/modules/audio_coding/main/include/audio_coding_module.h" |
| #include "webrtc/modules/audio_coding/main/acm2/acm_codec_database.h" |
| #include "webrtc/modules/audio_coding/main/acm2/acm_resampler.h" |
| #include "webrtc/modules/audio_coding/main/acm2/call_statistics.h" |
| #include "webrtc/modules/audio_coding/main/acm2/initial_delay_manager.h" |
| #include "webrtc/modules/audio_coding/neteq/include/neteq.h" |
| #include "webrtc/modules/interface/module_common_types.h" |
| #include "webrtc/typedefs.h" |
| |
| namespace webrtc { |
| |
| struct CodecInst; |
| class CriticalSectionWrapper; |
| class NetEq; |
| |
| namespace acm2 { |
| |
| class AcmReceiver { |
| public: |
| struct Decoder { |
| int acm_codec_id; |
| uint8_t payload_type; |
| // This field is meaningful for codecs where both mono and |
| // stereo versions are registered under the same ID. |
| int channels; |
| int sample_rate_hz; |
| }; |
| |
| // Constructor of the class |
| explicit AcmReceiver(const AudioCodingModule::Config& config); |
| |
| // Destructor of the class. |
| ~AcmReceiver(); |
| |
| // |
| // Inserts a payload with its associated RTP-header into NetEq. |
| // |
| // Input: |
| // - rtp_header : RTP header for the incoming payload containing |
| // information about payload type, sequence number, |
| // timestamp, SSRC and marker bit. |
| // - incoming_payload : Incoming audio payload. |
| // - length_payload : Length of incoming audio payload in bytes. |
| // |
| // Return value : 0 if OK. |
| // <0 if NetEq returned an error. |
| // |
| int InsertPacket(const WebRtcRTPHeader& rtp_header, |
| const uint8_t* incoming_payload, |
| size_t length_payload); |
| |
| // |
| // Asks NetEq for 10 milliseconds of decoded audio. |
| // |
| // Input: |
| // -desired_freq_hz : specifies the sampling rate [Hz] of the output |
| // audio. If set -1 indicates to resampling is |
| // is required and the audio returned at the |
| // sampling rate of the decoder. |
| // |
| // Output: |
| // -audio_frame : an audio frame were output data and |
| // associated parameters are written to. |
| // |
| // Return value : 0 if OK. |
| // -1 if NetEq returned an error. |
| // |
| int GetAudio(int desired_freq_hz, AudioFrame* audio_frame); |
| |
| // |
| // Adds a new codec to the NetEq codec database. |
| // |
| // Input: |
| // - acm_codec_id : ACM codec ID; -1 means external decoder. |
| // - payload_type : payload type. |
| // - sample_rate_hz : sample rate. |
| // - audio_decoder : pointer to a decoder object. If it's null, then |
| // NetEq will internally create a decoder object |
| // based on the value of |acm_codec_id| (which |
| // mustn't be -1). Otherwise, NetEq will use the |
| // given decoder for the given payload type. NetEq |
| // won't take ownership of the decoder; it's up to |
| // the caller to delete it when it's no longer |
| // needed. |
| // |
| // Providing an existing decoder object here is |
| // necessary for external decoders, but may also be |
| // used for built-in decoders if NetEq doesn't have |
| // all the info it needs to construct them properly |
| // (e.g. iSAC, where the decoder needs to be paired |
| // with an encoder). |
| // |
| // Return value : 0 if OK. |
| // <0 if NetEq returned an error. |
| // |
| int AddCodec(int acm_codec_id, |
| uint8_t payload_type, |
| int channels, |
| int sample_rate_hz, |
| AudioDecoder* audio_decoder); |
| |
| // |
| // Sets a minimum delay for packet buffer. The given delay is maintained, |
| // unless channel condition dictates a higher delay. |
| // |
| // Input: |
| // - delay_ms : minimum delay in milliseconds. |
| // |
| // Return value : 0 if OK. |
| // <0 if NetEq returned an error. |
| // |
| int SetMinimumDelay(int delay_ms); |
| |
| // |
| // Sets a maximum delay [ms] for the packet buffer. The target delay does not |
| // exceed the given value, even if channel condition requires so. |
| // |
| // Input: |
| // - delay_ms : maximum delay in milliseconds. |
| // |
| // Return value : 0 if OK. |
| // <0 if NetEq returned an error. |
| // |
| int SetMaximumDelay(int delay_ms); |
| |
| // |
| // Get least required delay computed based on channel conditions. Note that |
| // this is before applying any user-defined limits (specified by calling |
| // (SetMinimumDelay() and/or SetMaximumDelay()). |
| // |
| int LeastRequiredDelayMs() const; |
| |
| // |
| // Sets an initial delay of |delay_ms| milliseconds. This introduces a playout |
| // delay. Silence (zero signal) is played out until equivalent of |delay_ms| |
| // millisecond of audio is buffered. Then, NetEq maintains the delay. |
| // |
| // Input: |
| // - delay_ms : initial delay in milliseconds. |
| // |
| // Return value : 0 if OK. |
| // <0 if NetEq returned an error. |
| // |
| int SetInitialDelay(int delay_ms); |
| |
| // |
| // Resets the initial delay to zero. |
| // |
| void ResetInitialDelay(); |
| |
| // |
| // Get the current sampling frequency in Hz. |
| // |
| // Return value : Sampling frequency in Hz. |
| // |
| int current_sample_rate_hz() const; |
| |
| // |
| // Get the current network statistics from NetEq. |
| // |
| // Output: |
| // - statistics : The current network statistics. |
| // |
| void GetNetworkStatistics(NetworkStatistics* statistics); |
| |
| // |
| // Enable post-decoding VAD. |
| // |
| void EnableVad(); |
| |
| // |
| // Disable post-decoding VAD. |
| // |
| void DisableVad(); |
| |
| // |
| // Returns whether post-decoding VAD is enabled (true) or disabled (false). |
| // |
| bool vad_enabled() const { return vad_enabled_; } |
| |
| // |
| // Flushes the NetEq packet and speech buffers. |
| // |
| void FlushBuffers(); |
| |
| // |
| // Removes a payload-type from the NetEq codec database. |
| // |
| // Input: |
| // - payload_type : the payload-type to be removed. |
| // |
| // Return value : 0 if OK. |
| // -1 if an error occurred. |
| // |
| int RemoveCodec(uint8_t payload_type); |
| |
| // |
| // Remove all registered codecs. |
| // |
| int RemoveAllCodecs(); |
| |
| // |
| // Set ID. |
| // |
| void set_id(int id); // TODO(turajs): can be inline. |
| |
| // |
| // Gets the RTP timestamp of the last sample delivered by GetAudio(). |
| // Returns true if the RTP timestamp is valid, otherwise false. |
| // |
| bool GetPlayoutTimestamp(uint32_t* timestamp); |
| |
| // |
| // Return the index of the codec associated with the last non-CNG/non-DTMF |
| // received payload. If no non-CNG/non-DTMF payload is received -1 is |
| // returned. |
| // |
| int last_audio_codec_id() const; // TODO(turajs): can be inline. |
| |
| // |
| // Get the audio codec associated with the last non-CNG/non-DTMF received |
| // payload. If no non-CNG/non-DTMF packet is received -1 is returned, |
| // otherwise return 0. |
| // |
| int LastAudioCodec(CodecInst* codec) const; |
| |
| // |
| // Get a decoder given its registered payload-type. |
| // |
| // Input: |
| // -payload_type : the payload-type of the codec to be retrieved. |
| // |
| // Output: |
| // -codec : codec associated with the given payload-type. |
| // |
| // Return value : 0 if succeeded. |
| // -1 if failed, e.g. given payload-type is not |
| // registered. |
| // |
| int DecoderByPayloadType(uint8_t payload_type, |
| CodecInst* codec) const; |
| |
| // |
| // Enable NACK and set the maximum size of the NACK list. If NACK is already |
| // enabled then the maximum NACK list size is modified accordingly. |
| // |
| // Input: |
| // -max_nack_list_size : maximum NACK list size |
| // should be positive (none zero) and less than or |
| // equal to |Nack::kNackListSizeLimit| |
| // Return value |
| // : 0 if succeeded. |
| // -1 if failed |
| // |
| int EnableNack(size_t max_nack_list_size); |
| |
| // Disable NACK. |
| void DisableNack(); |
| |
| // |
| // Get a list of packets to be retransmitted. |
| // |
| // Input: |
| // -round_trip_time_ms : estimate of the round-trip-time (in milliseconds). |
| // Return value : list of packets to be retransmitted. |
| // |
| std::vector<uint16_t> GetNackList(int64_t round_trip_time_ms) const; |
| |
| // |
| // Get statistics of calls to GetAudio(). |
| void GetDecodingCallStatistics(AudioDecodingCallStats* stats) const; |
| |
| private: |
| bool GetSilence(int desired_sample_rate_hz, AudioFrame* frame) |
| EXCLUSIVE_LOCKS_REQUIRED(crit_sect_); |
| |
| int GetNumSyncPacketToInsert(uint16_t received_squence_number); |
| |
| const Decoder* RtpHeaderToDecoder(const RTPHeader& rtp_header, |
| const uint8_t* payload) const |
| EXCLUSIVE_LOCKS_REQUIRED(crit_sect_); |
| |
| uint32_t NowInTimestamp(int decoder_sampling_rate) const; |
| |
| void InsertStreamOfSyncPackets(InitialDelayManager::SyncStream* sync_stream); |
| |
| rtc::scoped_ptr<CriticalSectionWrapper> crit_sect_; |
| int id_; // TODO(henrik.lundin) Make const. |
| const Decoder* last_audio_decoder_ GUARDED_BY(crit_sect_); |
| AudioFrame::VADActivity previous_audio_activity_ GUARDED_BY(crit_sect_); |
| int current_sample_rate_hz_ GUARDED_BY(crit_sect_); |
| ACMResampler resampler_ GUARDED_BY(crit_sect_); |
| // Used in GetAudio, declared as member to avoid allocating every 10ms. |
| // TODO(henrik.lundin) Stack-allocate in GetAudio instead? |
| rtc::scoped_ptr<int16_t[]> audio_buffer_ GUARDED_BY(crit_sect_); |
| rtc::scoped_ptr<int16_t[]> last_audio_buffer_ GUARDED_BY(crit_sect_); |
| CallStatistics call_stats_ GUARDED_BY(crit_sect_); |
| NetEq* neteq_; |
| // Decoders map is keyed by payload type |
| std::map<uint8_t, Decoder> decoders_ GUARDED_BY(crit_sect_); |
| bool vad_enabled_; |
| Clock* clock_; // TODO(henrik.lundin) Make const if possible. |
| bool resampled_last_output_frame_ GUARDED_BY(crit_sect_); |
| |
| // Indicates if a non-zero initial delay is set, and the receiver is in |
| // AV-sync mode. |
| bool av_sync_; |
| rtc::scoped_ptr<InitialDelayManager> initial_delay_manager_; |
| |
| // The following are defined as members to avoid creating them in every |
| // iteration. |missing_packets_sync_stream_| is *ONLY* used in InsertPacket(). |
| // |late_packets_sync_stream_| is only used in GetAudio(). Both of these |
| // member variables are allocated only when we AV-sync is enabled, i.e. |
| // initial delay is set. |
| rtc::scoped_ptr<InitialDelayManager::SyncStream> missing_packets_sync_stream_; |
| rtc::scoped_ptr<InitialDelayManager::SyncStream> late_packets_sync_stream_; |
| }; |
| |
| } // namespace acm2 |
| |
| } // namespace webrtc |
| |
| #endif // WEBRTC_MODULES_AUDIO_CODING_MAIN_ACM2_ACM_RECEIVER_H_ |
