3 * Copyright 2004 Google Inc.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions are met:
8 * 1. Redistributions of source code must retain the above copyright notice,
9 * this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright notice,
11 * this list of conditions and the following disclaimer in the documentation
12 * and/or other materials provided with the distribution.
13 * 3. The name of the author may not be used to endorse or promote products
14 * derived from this software without specific prior written permission.
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
17 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
18 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
19 * EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
20 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
21 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
22 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
23 * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
24 * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
25 * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
28 #ifndef TALK_MEDIA_WEBRTCVOICEENGINE_H_
29 #define TALK_MEDIA_WEBRTCVOICEENGINE_H_
36 #include "talk/base/buffer.h"
37 #include "talk/base/byteorder.h"
38 #include "talk/base/logging.h"
39 #include "talk/base/scoped_ptr.h"
40 #include "talk/base/stream.h"
41 #include "talk/media/base/rtputils.h"
42 #include "talk/media/webrtc/webrtccommon.h"
43 #include "talk/media/webrtc/webrtcexport.h"
44 #include "talk/media/webrtc/webrtcvoe.h"
45 #include "talk/session/media/channel.h"
46 #include "webrtc/common.h"
47 #include "webrtc/modules/audio_coding/main/interface/audio_coding_module.h"
49 #if !defined(LIBPEERCONNECTION_LIB) && \
50 !defined(LIBPEERCONNECTION_IMPLEMENTATION)
51 #error "Bogus include."
57 // WebRtcSoundclipStream is an adapter object that allows a memory stream to be
58 // passed into WebRtc, and support looping.
59 class WebRtcSoundclipStream : public webrtc::InStream {
61 WebRtcSoundclipStream(const char* buf, size_t len)
62 : mem_(buf, len), loop_(true) {
64 void set_loop(bool loop) { loop_ = loop; }
65 virtual int Read(void* buf, int len);
69 talk_base::MemoryStream mem_;
73 // WebRtcMonitorStream is used to monitor a stream coming from WebRtc.
74 // For now we just dump the data.
75 class WebRtcMonitorStream : public webrtc::OutStream {
76 virtual bool Write(const void *buf, int len) {
81 class AudioDeviceModule;
83 class VoETraceWrapper;
86 class WebRtcSoundclipMedia;
87 class WebRtcVoiceMediaChannel;
89 // WebRtcVoiceEngine is a class to be used with CompositeMediaEngine.
90 // It uses the WebRtc VoiceEngine library for audio handling.
91 class WebRtcVoiceEngine
92 : public webrtc::VoiceEngineObserver,
93 public webrtc::TraceCallback,
94 public webrtc::VoEMediaProcess {
97 // Dependency injection for testing.
98 WebRtcVoiceEngine(VoEWrapper* voe_wrapper,
99 VoEWrapper* voe_wrapper_sc,
100 VoETraceWrapper* tracing);
101 ~WebRtcVoiceEngine();
102 bool Init(talk_base::Thread* worker_thread);
105 int GetCapabilities();
106 VoiceMediaChannel* CreateChannel();
108 SoundclipMedia* CreateSoundclip();
110 AudioOptions GetOptions() const { return options_; }
111 bool SetOptions(const AudioOptions& options);
112 // Overrides, when set, take precedence over the options on a
113 // per-option basis. For example, if AGC is set in options and AEC
114 // is set in overrides, AGC and AEC will be both be set. Overrides
115 // can also turn off options. For example, if AGC is set to "on" in
116 // options and AGC is set to "off" in overrides, the result is that
117 // AGC will be off until different overrides are applied or until
118 // the overrides are cleared. Only one set of overrides is present
119 // at a time (they do not "stack"). And when the overrides are
120 // cleared, the media engine's state reverts back to the options set
121 // via SetOptions. This allows us to have both "persistent options"
122 // (the normal options) and "temporary options" (overrides).
123 bool SetOptionOverrides(const AudioOptions& options);
124 bool ClearOptionOverrides();
125 bool SetDelayOffset(int offset);
126 bool SetDevices(const Device* in_device, const Device* out_device);
127 bool GetOutputVolume(int* level);
128 bool SetOutputVolume(int level);
130 bool SetLocalMonitor(bool enable);
132 const std::vector<AudioCodec>& codecs();
133 bool FindCodec(const AudioCodec& codec);
134 bool FindWebRtcCodec(const AudioCodec& codec, webrtc::CodecInst* gcodec);
136 const std::vector<RtpHeaderExtension>& rtp_header_extensions() const;
138 void SetLogging(int min_sev, const char* filter);
140 bool RegisterProcessor(uint32 ssrc,
141 VoiceProcessor* voice_processor,
142 MediaProcessorDirection direction);
143 bool UnregisterProcessor(uint32 ssrc,
144 VoiceProcessor* voice_processor,
145 MediaProcessorDirection direction);
147 // Method from webrtc::VoEMediaProcess
148 virtual void Process(int channel,
149 webrtc::ProcessingTypes type,
155 // For tracking WebRtc channels. Needed because we have to pause them
156 // all when switching devices.
157 // May only be called by WebRtcVoiceMediaChannel.
158 void RegisterChannel(WebRtcVoiceMediaChannel *channel);
159 void UnregisterChannel(WebRtcVoiceMediaChannel *channel);
161 // May only be called by WebRtcSoundclipMedia.
162 void RegisterSoundclip(WebRtcSoundclipMedia *channel);
163 void UnregisterSoundclip(WebRtcSoundclipMedia *channel);
165 // Called by WebRtcVoiceMediaChannel to set a gain offset from
166 // the default AGC target level.
167 bool AdjustAgcLevel(int delta);
169 VoEWrapper* voe() { return voe_wrapper_.get(); }
170 VoEWrapper* voe_sc() { return voe_wrapper_sc_.get(); }
171 int GetLastEngineError();
173 // Set the external ADMs. This can only be called before Init.
174 bool SetAudioDeviceModule(webrtc::AudioDeviceModule* adm,
175 webrtc::AudioDeviceModule* adm_sc);
177 // Starts AEC dump using existing file.
178 bool StartAecDump(talk_base::PlatformFile file);
180 // Check whether the supplied trace should be ignored.
181 bool ShouldIgnoreTrace(const std::string& trace);
183 // Create a VoiceEngine Channel.
184 int CreateMediaVoiceChannel();
185 int CreateSoundclipVoiceChannel();
188 typedef std::vector<WebRtcSoundclipMedia *> SoundclipList;
189 typedef std::vector<WebRtcVoiceMediaChannel *> ChannelList;
191 signal3<uint32, MediaProcessorDirection, AudioFrame*> FrameSignal;
194 void ConstructCodecs();
196 bool EnsureSoundclipEngineInit();
197 void SetTraceFilter(int filter);
198 void SetTraceOptions(const std::string& options);
199 // Applies either options or overrides. Every option that is "set"
200 // will be applied. Every option not "set" will be ignored. This
201 // allows us to selectively turn on and off different options easily
203 bool ApplyOptions(const AudioOptions& options);
204 // Configure for using ACM2, if |enable| is true, otherwise configure for
206 void EnableExperimentalAcm(bool enable);
207 virtual void Print(webrtc::TraceLevel level, const char* trace, int length);
208 virtual void CallbackOnError(int channel, int errCode);
209 // Given the device type, name, and id, find device id. Return true and
210 // set the output parameter rtc_id if successful.
211 bool FindWebRtcAudioDeviceId(
212 bool is_input, const std::string& dev_name, int dev_id, int* rtc_id);
213 bool FindChannelAndSsrc(int channel_num,
214 WebRtcVoiceMediaChannel** channel,
216 bool FindChannelNumFromSsrc(uint32 ssrc,
217 MediaProcessorDirection direction,
219 bool ChangeLocalMonitor(bool enable);
220 bool PauseLocalMonitor();
221 bool ResumeLocalMonitor();
223 bool UnregisterProcessorChannel(MediaProcessorDirection channel_direction,
225 VoiceProcessor* voice_processor,
226 MediaProcessorDirection processor_direction);
228 void StartAecDump(const std::string& filename);
230 int CreateVoiceChannel(VoEWrapper* voe);
232 // When a voice processor registers with the engine, it is connected
233 // to either the Rx or Tx signals, based on the direction parameter.
234 // SignalXXMediaFrame will be invoked for every audio packet.
235 FrameSignal SignalRxMediaFrame;
236 FrameSignal SignalTxMediaFrame;
238 static const int kDefaultLogSeverity = talk_base::LS_WARNING;
240 // The primary instance of WebRtc VoiceEngine.
241 talk_base::scoped_ptr<VoEWrapper> voe_wrapper_;
242 // A secondary instance, for playing out soundclips (on the 'ring' device).
243 talk_base::scoped_ptr<VoEWrapper> voe_wrapper_sc_;
244 bool voe_wrapper_sc_initialized_;
245 talk_base::scoped_ptr<VoETraceWrapper> tracing_;
246 // The external audio device manager
247 webrtc::AudioDeviceModule* adm_;
248 webrtc::AudioDeviceModule* adm_sc_;
250 std::string log_options_;
251 bool is_dumping_aec_;
252 std::vector<AudioCodec> codecs_;
253 std::vector<RtpHeaderExtension> rtp_header_extensions_;
254 bool desired_local_monitor_enable_;
255 talk_base::scoped_ptr<WebRtcMonitorStream> monitor_;
256 SoundclipList soundclips_;
257 ChannelList channels_;
258 // channels_ can be read from WebRtc callback thread. We need a lock on that
259 // callback as well as the RegisterChannel/UnregisterChannel.
260 talk_base::CriticalSection channels_cs_;
261 webrtc::AgcConfig default_agc_config_;
263 webrtc::Config voe_config_;
264 bool use_experimental_acm_;
267 // See SetOptions and SetOptionOverrides for a description of the
268 // difference between options and overrides.
269 // options_ are the base options, which combined with the
270 // option_overrides_, create the current options being used.
271 // options_ is stored so that when option_overrides_ is cleared, we
272 // can restore the options_ without the option_overrides.
273 AudioOptions options_;
274 AudioOptions option_overrides_;
276 // When the media processor registers with the engine, the ssrc is cached
277 // here so that a look up need not be made when the callback is invoked.
278 // This is necessary because the lookup results in mux_channels_cs lock being
279 // held and if a remote participant leaves the hangout at the same time
280 // we hit a deadlock.
281 uint32 tx_processor_ssrc_;
282 uint32 rx_processor_ssrc_;
284 talk_base::CriticalSection signal_media_critical_;
287 // WebRtcMediaChannel is a class that implements the common WebRtc channel
289 template <class T, class E>
290 class WebRtcMediaChannel : public T, public webrtc::Transport {
292 WebRtcMediaChannel(E *engine, int channel)
293 : engine_(engine), voe_channel_(channel) {}
294 E *engine() { return engine_; }
295 int voe_channel() const { return voe_channel_; }
296 bool valid() const { return voe_channel_ != -1; }
299 // implements Transport interface
300 virtual int SendPacket(int channel, const void *data, int len) {
301 talk_base::Buffer packet(data, len, kMaxRtpPacketLen);
302 if (!T::SendPacket(&packet)) {
308 virtual int SendRTCPPacket(int channel, const void *data, int len) {
309 talk_base::Buffer packet(data, len, kMaxRtpPacketLen);
310 return T::SendRtcp(&packet) ? len : -1;
318 // WebRtcVoiceMediaChannel is an implementation of VoiceMediaChannel that uses
319 // WebRtc Voice Engine.
320 class WebRtcVoiceMediaChannel
321 : public WebRtcMediaChannel<VoiceMediaChannel, WebRtcVoiceEngine> {
323 explicit WebRtcVoiceMediaChannel(WebRtcVoiceEngine *engine);
324 virtual ~WebRtcVoiceMediaChannel();
325 virtual bool SetOptions(const AudioOptions& options);
326 virtual bool GetOptions(AudioOptions* options) const {
330 virtual bool SetRecvCodecs(const std::vector<AudioCodec> &codecs);
331 virtual bool SetSendCodecs(const std::vector<AudioCodec> &codecs);
332 virtual bool SetRecvRtpHeaderExtensions(
333 const std::vector<RtpHeaderExtension>& extensions);
334 virtual bool SetSendRtpHeaderExtensions(
335 const std::vector<RtpHeaderExtension>& extensions);
336 virtual bool SetPlayout(bool playout);
338 bool ResumePlayout();
339 virtual bool SetSend(SendFlags send);
342 virtual bool AddSendStream(const StreamParams& sp);
343 virtual bool RemoveSendStream(uint32 ssrc);
344 virtual bool AddRecvStream(const StreamParams& sp);
345 virtual bool RemoveRecvStream(uint32 ssrc);
346 virtual bool SetRemoteRenderer(uint32 ssrc, AudioRenderer* renderer);
347 virtual bool SetLocalRenderer(uint32 ssrc, AudioRenderer* renderer);
348 virtual bool GetActiveStreams(AudioInfo::StreamList* actives);
349 virtual int GetOutputLevel();
350 virtual int GetTimeSinceLastTyping();
351 virtual void SetTypingDetectionParameters(int time_window,
352 int cost_per_typing, int reporting_threshold, int penalty_decay,
353 int type_event_delay);
354 virtual bool SetOutputScaling(uint32 ssrc, double left, double right);
355 virtual bool GetOutputScaling(uint32 ssrc, double* left, double* right);
357 virtual bool SetRingbackTone(const char *buf, int len);
358 virtual bool PlayRingbackTone(uint32 ssrc, bool play, bool loop);
359 virtual bool CanInsertDtmf();
360 virtual bool InsertDtmf(uint32 ssrc, int event, int duration, int flags);
362 virtual void OnPacketReceived(talk_base::Buffer* packet,
363 const talk_base::PacketTime& packet_time);
364 virtual void OnRtcpReceived(talk_base::Buffer* packet,
365 const talk_base::PacketTime& packet_time);
366 virtual void OnReadyToSend(bool ready) {}
367 virtual bool MuteStream(uint32 ssrc, bool on);
368 virtual bool SetStartSendBandwidth(int bps);
369 virtual bool SetMaxSendBandwidth(int bps);
370 virtual bool GetStats(VoiceMediaInfo* info);
371 // Gets last reported error from WebRtc voice engine. This should be only
372 // called in response a failure.
373 virtual void GetLastMediaError(uint32* ssrc,
374 VoiceMediaChannel::Error* error);
375 bool FindSsrc(int channel_num, uint32* ssrc);
376 void OnError(uint32 ssrc, int error);
378 bool sending() const { return send_ != SEND_NOTHING; }
379 int GetReceiveChannelNum(uint32 ssrc);
380 int GetSendChannelNum(uint32 ssrc);
383 int GetLastEngineError() { return engine()->GetLastEngineError(); }
384 int GetOutputLevel(int channel);
385 bool GetRedSendCodec(const AudioCodec& red_codec,
386 const std::vector<AudioCodec>& all_codecs,
387 webrtc::CodecInst* send_codec);
388 bool EnableRtcp(int channel);
389 bool ResetRecvCodecs(int channel);
390 bool SetPlayout(int channel, bool playout);
391 static uint32 ParseSsrc(const void* data, size_t len, bool rtcp);
392 static Error WebRtcErrorToChannelError(int err_code);
395 class WebRtcVoiceChannelRenderer;
396 // Map of ssrc to WebRtcVoiceChannelRenderer object. A new object of
397 // WebRtcVoiceChannelRenderer will be created for every new stream and
398 // will be destroyed when the stream goes away.
399 typedef std::map<uint32, WebRtcVoiceChannelRenderer*> ChannelMap;
401 void SetNack(int channel, bool nack_enabled);
402 void SetNack(const ChannelMap& channels, bool nack_enabled);
403 bool SetSendCodec(const webrtc::CodecInst& send_codec);
404 bool SetSendCodec(int channel, const webrtc::CodecInst& send_codec);
405 bool ChangePlayout(bool playout);
406 bool ChangeSend(SendFlags send);
407 bool ChangeSend(int channel, SendFlags send);
408 void ConfigureSendChannel(int channel);
409 bool ConfigureRecvChannel(int channel);
410 bool DeleteChannel(int channel);
411 bool InConferenceMode() const {
412 return options_.conference_mode.GetWithDefaultIfUnset(false);
414 bool IsDefaultChannel(int channel_id) const {
415 return channel_id == voe_channel();
417 bool SetSendCodecs(int channel, const std::vector<AudioCodec>& codecs);
418 bool SetSendBandwidthInternal(int bps);
420 talk_base::scoped_ptr<WebRtcSoundclipStream> ringback_tone_;
421 std::set<int> ringback_channels_; // channels playing ringback
422 std::vector<AudioCodec> recv_codecs_;
423 std::vector<AudioCodec> send_codecs_;
424 talk_base::scoped_ptr<webrtc::CodecInst> send_codec_;
425 bool send_bw_setting_;
427 AudioOptions options_;
429 bool desired_playout_;
432 bool typing_noise_detected_;
433 SendFlags desired_send_;
436 // send_channels_ contains the channels which are being used for sending.
437 // When the default channel (voe_channel) is used for sending, it is
438 // contained in send_channels_, otherwise not.
439 ChannelMap send_channels_;
440 uint32 default_receive_ssrc_;
441 // Note the default channel (voe_channel()) can reside in both
442 // receive_channels_ and send_channels_ in non-conference mode and in that
443 // case it will only be there if a non-zero default_receive_ssrc_ is set.
444 ChannelMap receive_channels_; // for multiple sources
445 // receive_channels_ can be read from WebRtc callback thread. Access from
446 // the WebRtc thread must be synchronized with edits on the worker thread.
447 // Reads on the worker thread are ok.
449 // Do not lock this on the VoE media processor thread; potential for deadlock
451 mutable talk_base::CriticalSection receive_channels_cs_;
454 } // namespace cricket
456 #endif // TALK_MEDIA_WEBRTCVOICEENGINE_H_