more work on Voice

5 files changed, 214 insertions, 44 deletions

diff --git a/source/CMakeLists.txt b/source/CMakeLists.txt
index d3d60a3..4f44bc2 100644
--- a/source/CMakeLists.txt
+++ b/source/CMakeLists.txt
@@ -437,6 +437,7 @@ SET (STAR_EXT_LIBS ${STAR_EXT_LIBS}
     ${FREETYPE_LIBRARY}
     ${PNG_LIBRARY}
     ${ZLIB_LIBRARY}
+    "extern/opus/opus"
   )
 
 IF (STAR_BUILD_GUI)
diff --git a/source/extern/CMakeLists.txt b/source/extern/CMakeLists.txt
index e356c95..5ea7420 100644
--- a/source/extern/CMakeLists.txt
+++ b/source/extern/CMakeLists.txt
@@ -79,4 +79,4 @@ SET (star_extern_SOURCES
   )
 
 ADD_LIBRARY (star_extern OBJECT ${star_extern_SOURCES} ${star_extern_HEADERS})
-TARGET_LINK_LIBRARIES(star_extern opus)
\ No newline at end of file
+TARGET_LINK_LIBRARIES(star_extern PUBLIC opus)
\ No newline at end of file
diff --git a/source/frontend/StarMainMixer.cpp b/source/frontend/StarMainMixer.cpp
index 59b6b52..6d68ea2 100644
--- a/source/frontend/StarMainMixer.cpp
+++ b/source/frontend/StarMainMixer.cpp
@@ -7,6 +7,7 @@
 #include "StarAssets.hpp"
 #include "StarWorldClient.hpp"
 #include "StarWorldPainter.hpp"
+#include "StarVoice.hpp"
 
 namespace Star {
 
@@ -80,34 +81,39 @@ void MainMixer::update(bool muteSfx, bool muteMusic) {
     auto cameraPos = m_worldPainter->camera().centerWorldPosition();
     auto worldGeometry = currentWorld->geometry();
 
-    m_mixer->update([&](unsigned channel, Vec2F pos, float rangeMultiplier) {
-        Vec2F playerDiff = worldGeometry.diff(pos, playerPos);
-        Vec2F cameraDiff = worldGeometry.diff(pos, cameraPos);
-        float playerMagSq = playerDiff.magnitudeSquared();
-        float cameraMagSq = cameraDiff.magnitudeSquared();
-
-        Vec2F diff;
-        float diffMagnitude;
-        if (playerMagSq < cameraMagSq) {
-          diff = playerDiff;
-          diffMagnitude = sqrt(playerMagSq);
-        }
-        else {
-          diff = cameraDiff;
-          diffMagnitude = sqrt(cameraMagSq);
-        }
+    Mixer::PositionalAttenuationFunction attenuationFunction = [&](unsigned channel, Vec2F pos, float rangeMultiplier) {
+      Vec2F playerDiff = worldGeometry.diff(pos, playerPos);
+      Vec2F cameraDiff = worldGeometry.diff(pos, cameraPos);
+      float playerMagSq = playerDiff.magnitudeSquared();
+      float cameraMagSq = cameraDiff.magnitudeSquared();
+
+      Vec2F diff;
+      float diffMagnitude;
+      if (playerMagSq < cameraMagSq) {
+        diff = playerDiff;
+        diffMagnitude = sqrt(playerMagSq);
+      }
+      else {
+        diff = cameraDiff;
+        diffMagnitude = sqrt(cameraMagSq);
+      }
 
-        if (diffMagnitude == 0.0f)
-          return 0.0f;
+      if (diffMagnitude == 0.0f)
+        return 0.0f;
 
-        Vec2F diffNorm = diff / diffMagnitude;
+      Vec2F diffNorm = diff / diffMagnitude;
 
-        float stereoIncidence = channel == 0 ? -diffNorm[0] : diffNorm[0];
+      float stereoIncidence = channel == 0 ? -diffNorm[0] : diffNorm[0];
+
+      float maxDistance = baseMaxDistance * rangeMultiplier * lerp((stereoIncidence + 1.0f) / 2.0f, stereoAdjustmentRange[0], stereoAdjustmentRange[1]);
+
+      return pow(clamp(diffMagnitude / maxDistance, 0.0f, 1.0f), 1.0f / attenuationGamma);
+    };
 
-        float maxDistance = baseMaxDistance * rangeMultiplier * lerp((stereoIncidence + 1.0f) / 2.0f, stereoAdjustmentRange[0], stereoAdjustmentRange[1]);
+    if (Voice* voice = Voice::singletonPtr())
+      voice->update(attenuationFunction);
 
-        return pow(clamp(diffMagnitude / maxDistance, 0.0f, 1.0f), 1.0f / attenuationGamma);
-      });
+    m_mixer->update(attenuationFunction);
 
   } else {
     if (m_mixer->hasEffect("lowpass"))
diff --git a/source/game/StarVoice.cpp b/source/game/StarVoice.cpp
index 1c949c1..081e2a6 100644
--- a/source/game/StarVoice.cpp
+++ b/source/game/StarVoice.cpp
@@ -1,11 +1,35 @@
 #include "StarVoice.hpp"
+#include "StarFormat.hpp"
+#include "opus/include/opus.h"
+
+#include "SDL.h"
 
 namespace Star {
 
-STAR_EXCEPTION(VoiceException, StarException);
+EnumMap<VoiceTriggerMode> const VoiceTriggerModeNames{
+  {VoiceTriggerMode::VoiceActivity, "VoiceActivity"},
+  {VoiceTriggerMode::PushToTalk, "PushToTalk"}
+};
 
-void Voice::mix(int16_t* buffer, size_t frames, unsigned channels) {
+EnumMap<VoiceChannelMode> const VoiceChannelModeNames{
+  {VoiceChannelMode::Mono, "Mono"},
+  {VoiceChannelMode::Stereo, "Stereo"}
+};
+
+static SDL_AudioDeviceID sdlInputDevice = 0;
+
+constexpr int VOICE_SAMPLE_RATE = 48000;
+constexpr int VOICE_FRAME_SIZE = 960;
 
+constexpr int VOICE_MAX_FRAME_SIZE = 6 * VOICE_FRAME_SIZE;
+constexpr int VOICE_MAX_PACKET_SIZE = 3 * 1276;
+
+constexpr uint16_t VOICE_VERSION = 1;
+
+Voice::Speaker::Speaker(SpeakerId id)
+  : decoderMono  (createDecoder(1), opus_decoder_destroy)
+  , decoderStereo(createDecoder(2), opus_decoder_destroy) {
+  speakerId = id;
 }
 
 Voice* Voice::s_singleton;
@@ -21,10 +45,15 @@ Voice& Voice::singleton() {
     return *s_singleton;
 }
 
-Voice::Voice() {
+Voice::Voice() : m_encoder(nullptr, opus_encoder_destroy) {
   if (s_singleton)
     throw VoiceException("Singleton Voice has been constructed twice");
 
+  m_clientSpeaker = make_shared<Speaker>(m_speakerId);
+  m_triggerMode = VoiceTriggerMode::PushToTalk;
+  m_channelMode = VoiceChannelMode::Mono;
+
+  resetEncoder();
   s_singleton = this;
 }
 
@@ -32,4 +61,71 @@ Voice::~Voice() {
   s_singleton = nullptr;
 }
 
+void Voice::load(Json const& config) {
+  // do stuff
+}
+Json Voice::save() const {
+  return JsonObject{};
+}
+
+Voice::SpeakerPtr Voice::setLocalSpeaker(SpeakerId speakerId) {
+  if (m_speakers.contains(m_speakerId))
+    m_speakers.remove(m_speakerId);
+
+  m_clientSpeaker->speakerId = m_speakerId = speakerId;
+  return m_speakers.insert(m_speakerId, m_clientSpeaker).first->second;
+}
+
+Voice::SpeakerPtr Voice::speaker(SpeakerId speakerId) {
+  if (m_speakerId == speakerId)
+    return m_clientSpeaker;
+  else {
+    if (SpeakerPtr const* ptr = m_speakers.ptr(speakerId))
+      return *ptr;
+    else
+      return m_speakers.emplace(speakerId, make_shared<Speaker>(speakerId)).first->second;
+  }
+}
+
+void Voice::mix(int16_t* buffer, size_t frames, unsigned channels) {
+
+}
+
+void Voice::update(PositionalAttenuationFunction positionalAttenuationFunction) {
+  if (positionalAttenuationFunction) {
+    for (auto& entry : m_speakers) {
+      if (SpeakerPtr& speaker = entry.second) {
+        speaker->channelVolumes = {
+          positionalAttenuationFunction(0, speaker->position, 1.0f),
+          positionalAttenuationFunction(1, speaker->position, 1.0f)
+        };
+      }
+    }
+  }
+}
+
+OpusDecoder* Voice::createDecoder(int channels) {
+  int error;
+  OpusDecoder* decoder = opus_decoder_create(VOICE_SAMPLE_RATE, channels, &error);
+  if (error != OPUS_OK)
+    throw VoiceException::format("Could not create decoder: {}", opus_strerror(error));
+  else
+    return decoder;
+}
+
+OpusEncoder* Voice::createEncoder(int channels) {
+  int error;
+  OpusEncoder* encoder = opus_encoder_create(VOICE_SAMPLE_RATE, channels, OPUS_APPLICATION_AUDIO, &error);
+  if (error != OPUS_OK)
+    throw VoiceException::format("Could not create encoder: {}", opus_strerror(error));
+  else
+    return encoder;
+}
+
+void Voice::resetEncoder() {
+  int channels = encoderChannels();
+  m_encoder.reset(createEncoder(channels));
+  opus_encoder_ctl(m_encoder.get(), OPUS_SET_BITRATE(channels == 2 ? 50000 : 24000));
+}
+
 }
\ No newline at end of file
diff --git a/source/game/StarVoice.hpp b/source/game/StarVoice.hpp
index f102889..e870439 100644
--- a/source/game/StarVoice.hpp
+++ b/source/game/StarVoice.hpp
@@ -1,31 +1,98 @@
 #ifndef STAR_VOICE_HPP
 #define STAR_VOICE_HPP
-#include "StarString.hpp"
+#include "StarJson.hpp"
+#include "StarBiMap.hpp"
+#include "StarGameTypes.hpp"
+#include "StarException.hpp"
+
+struct OpusDecoder;
+typedef std::unique_ptr<OpusDecoder, void(*)(OpusDecoder*)> OpusDecoderPtr;
+struct OpusEncoder;
+typedef std::unique_ptr<OpusEncoder, void(*)(OpusEncoder*)> OpusEncoderPtr;
 
 namespace Star {
 
-  STAR_CLASS(Voice);
+STAR_EXCEPTION(VoiceException, StarException);
+
+enum class VoiceTriggerMode : uint8_t { VoiceActivity, PushToTalk };
+extern EnumMap<VoiceTriggerMode> const VoiceTriggerModeNames;
+
+enum class VoiceChannelMode: uint8_t { Mono = 1, Stereo = 2 };
+extern EnumMap<VoiceChannelMode> const VoiceChannelModeNames;
 
-  class Voice {
-  public:
-    void mix(int16_t* buffer, size_t frames, unsigned channels);
+STAR_CLASS(Voice);
 
-    // Get pointer to the singleton Voice instance, if it exists.  Otherwise,
-    // returns nullptr.
-    static Voice* singletonPtr();
+class Voice {
+public:
+  // Individual speakers are represented by their connection ID.
+  typedef ConnectionId SpeakerId;
 
-    // Gets reference to Voice singleton, throws VoiceException if root
-    // is not initialized.
-    static Voice& singleton();
+  struct Speaker {
+    SpeakerId speakerId = 0;
+    EntityId entityId = 0;
+    Vec2F position = Vec2F();
+    String name = "Unnamed";
 
-    Voice();
-    ~Voice();
+    OpusDecoderPtr decoderMono;
+    OpusDecoderPtr decoderStereo;
 
-    Voice(Voice const&) = delete;
-    Voice& operator=(Voice const&) = delete;
-  private:
-    static Voice* s_singleton;
+    atomic<bool> active = false;
+    atomic<float> currentLoudness = 0.0f;
+    atomic<Array<float, 2>> channelVolumes = Array<float, 2>::filled(1.0f);
+
+    Speaker(SpeakerId speakerId);
   };
+
+  typedef std::shared_ptr<Speaker> SpeakerPtr;
+
+  // Get pointer to the singleton Voice instance, if it exists.  Otherwise,
+  // returns nullptr.
+  static Voice* singletonPtr();
+
+  // Gets reference to Voice singleton, throws VoiceException if root
+  // is not initialized.
+  static Voice& singleton();
+
+  Voice();
+  ~Voice();
+
+  Voice(Voice const&) = delete;
+  Voice& operator=(Voice const&) = delete;
+
+  void load(Json const& config);
+  Json save() const;
+  
+  // Sets the local speaker ID and returns the local speaker. Must be called upon loading into a world.
+  SpeakerPtr setLocalSpeaker(SpeakerId speakerId);
+  SpeakerPtr speaker(SpeakerId speakerId);
+
+  // Called to mix voice audio with the game.
+  void mix(int16_t* buffer, size_t frames, unsigned channels);
+
+  typedef function<float(unsigned, Vec2F, float)> PositionalAttenuationFunction;
+  void update(PositionalAttenuationFunction positionalAttenuationFunction = {});
+
+  inline int encoderChannels() const {
+    return m_channelMode == VoiceChannelMode::Mono ? 1 : 2;
+  }
+private:
+  static Voice* s_singleton;
+
+  static OpusDecoder* createDecoder(int channels);
+  static OpusEncoder* createEncoder(int channels);
+  void resetEncoder();
+
+  SpeakerId m_speakerId = 0;
+  SpeakerPtr m_clientSpeaker;
+  HashMap<SpeakerId, SpeakerPtr> m_speakers;
+
+  HashSet<SpeakerPtr> m_activeSpeakers;
+
+  OpusEncoderPtr m_encoder;
+
+  VoiceTriggerMode m_triggerMode;
+  VoiceChannelMode m_channelMode;
+};
   
 }