feat(audio): custom surround-params (#2424)

src/audio.cpp

@@ -34,6 +34,8 @@ namespace audio {
   start_audio_control(audio_ctx_t &ctx);
   static void
   stop_audio_control(audio_ctx_t &);
+  static void
+  apply_surround_params(opus_stream_config_t &stream, const stream_params_t &params);
 
   int
   map_stream(int channels, bool quality);
@@ -98,28 +100,31 @@ namespace audio {
   void
   encodeThread(sample_queue_t samples, config_t config, void *channel_data) {
     auto packets = mail::man->queue<packet_t>(mail::audio_packets);
-    auto stream = &stream_configs[map_stream(config.channels, config.flags[config_t::HIGH_QUALITY])];
+    auto stream = stream_configs[map_stream(config.channels, config.flags[config_t::HIGH_QUALITY])];
+    if (config.flags[config_t::CUSTOM_SURROUND_PARAMS]) {
+      apply_surround_params(stream, config.customStreamParams);
+    }
 
     // Encoding takes place on this thread
     platf::adjust_thread_priority(platf::thread_priority_e::high);
 
     opus_t opus { opus_multistream_encoder_create(
-      stream->sampleRate,
+      stream.sampleRate,
-      stream->channelCount,
+      stream.channelCount,
-      stream->streams,
+      stream.streams,
-      stream->coupledStreams,
+      stream.coupledStreams,
-      stream->mapping,
+      stream.mapping,
       OPUS_APPLICATION_RESTRICTED_LOWDELAY,
       nullptr) };
 
-    opus_multistream_encoder_ctl(opus.get(), OPUS_SET_BITRATE(stream->bitrate));
+    opus_multistream_encoder_ctl(opus.get(), OPUS_SET_BITRATE(stream.bitrate));
     opus_multistream_encoder_ctl(opus.get(), OPUS_SET_VBR(0));
 
-    BOOST_LOG(info) << "Opus initialized: "sv << stream->sampleRate / 1000 << " kHz, "sv
+    BOOST_LOG(info) << "Opus initialized: "sv << stream.sampleRate / 1000 << " kHz, "sv
-                    << stream->channelCount << " channels, "sv
+                    << stream.channelCount << " channels, "sv
-                    << stream->bitrate / 1000 << " kbps (total), LOWDELAY"sv;
+                    << stream.bitrate / 1000 << " kbps (total), LOWDELAY"sv;
 
-    auto frame_size = config.packetDuration * stream->sampleRate / 1000;
+    auto frame_size = config.packetDuration * stream.sampleRate / 1000;
     while (auto sample = samples->pop()) {
       buffer_t packet { 1400 };
 
@@ -139,7 +144,10 @@ namespace audio {
   void
   capture(safe::mail_t mail, config_t config, void *channel_data) {
     auto shutdown_event = mail->event<bool>(mail::shutdown);
-    auto stream = &stream_configs[map_stream(config.channels, config.flags[config_t::HIGH_QUALITY])];
+    auto stream = stream_configs[map_stream(config.channels, config.flags[config_t::HIGH_QUALITY])];
+    if (config.flags[config_t::CUSTOM_SURROUND_PARAMS]) {
+      apply_surround_params(stream, config.customStreamParams);
+    }
 
     auto ref = control_shared.ref();
     if (!ref) {
@@ -171,7 +179,7 @@ namespace audio {
     // Prefer the virtual sink if host playback is disabled or there's no other sink
     if (ref->sink.null && (!config.flags[config_t::HOST_AUDIO] || sink->empty())) {
       auto &null = *ref->sink.null;
-      switch (stream->channelCount) {
+      switch (stream.channelCount) {
         case 2:
           sink = &null.stereo;
           break;
@@ -195,8 +203,8 @@ namespace audio {
       }
     }
 
-    auto frame_size = config.packetDuration * stream->sampleRate / 1000;
+    auto frame_size = config.packetDuration * stream.sampleRate / 1000;
-    auto mic = control->microphone(stream->mapping, stream->channelCount, stream->sampleRate, frame_size);
+    auto mic = control->microphone(stream.mapping, stream.channelCount, stream.sampleRate, frame_size);
     if (!mic) {
       return;
     }
@@ -217,7 +225,7 @@ namespace audio {
       shutdown_event->view();
     });
 
-    int samples_per_frame = frame_size * stream->channelCount;
+    int samples_per_frame = frame_size * stream.channelCount;
 
     while (!shutdown_event->peek()) {
       std::vector<std::int16_t> sample_buffer;
@@ -233,7 +241,7 @@ namespace audio {
           BOOST_LOG(info) << "Reinitializing audio capture"sv;
           mic.reset();
           do {
-            mic = control->microphone(stream->mapping, stream->channelCount, stream->sampleRate, frame_size);
+            mic = control->microphone(stream.mapping, stream.channelCount, stream.sampleRate, frame_size);
             if (!mic) {
               BOOST_LOG(warning) << "Couldn't re-initialize audio input"sv;
             }
@@ -303,4 +311,12 @@ namespace audio {
       ctx.control->set_sink(sink);
     }
   }
+
+  void
+  apply_surround_params(opus_stream_config_t &stream, const stream_params_t &params) {
+    stream.channelCount = params.channelCount;
+    stream.streams = params.streams;
+    stream.coupledStreams = params.coupledStreams;
+    stream.mapping = params.mapping;
+  }
 }  // namespace audio

src/audio.h

@@ -26,12 +26,20 @@ namespace audio {
     int bitrate;
   };
 
+  struct stream_params_t {
+    int channelCount;
+    int streams;
+    int coupledStreams;
+    std::uint8_t mapping[8];
+  };
+
   extern opus_stream_config_t stream_configs[MAX_STREAM_CONFIG];
 
   struct config_t {
     enum flags_e : int {
       HIGH_QUALITY,
       HOST_AUDIO,
+      CUSTOM_SURROUND_PARAMS,
       MAX_FLAGS
     };
 
@@ -39,6 +47,8 @@ namespace audio {
     int channels;
     int mask;
 
+    stream_params_t customStreamParams;
+
     std::bitset<MAX_FLAGS> flags;
   };
 

src/nvhttp.cpp

@@ -330,6 +330,7 @@ namespace nvhttp {
     launch_session->appid = util::from_view(get_arg(args, "appid", "unknown"));
     launch_session->enable_sops = util::from_view(get_arg(args, "sops", "0"));
     launch_session->surround_info = util::from_view(get_arg(args, "surroundAudioInfo", "196610"));
+    launch_session->surround_params = (get_arg(args, "surroundParams", ""));
     launch_session->gcmap = util::from_view(get_arg(args, "gcmap", "0"));
     launch_session->enable_hdr = util::from_view(get_arg(args, "hdrMode", "0"));
 

src/process.cpp

@@ -185,6 +185,7 @@ namespace proc {
         _env["SUNSHINE_CLIENT_AUDIO_CONFIGURATION"] = "7.1";
         break;
     }
+    _env["SUNSHINE_CLIENT_AUDIO_SURROUND_PARAMS"] = launch_session->surround_params;
 
     if (!_app.output.empty() && _app.output != "null"sv) {
 #ifdef _WIN32

src/rtsp.cpp

@@ -819,6 +819,12 @@ namespace rtsp_stream {
       ss << "a=rtpmap:98 AV1/90000"sv << std::endl;
     }
 
+    if (!session.surround_params.empty()) {
+      // If we have our own surround parameters, advertise them twice first
+      ss << "a=fmtp:97 surround-params="sv << session.surround_params << std::endl;
+      ss << "a=fmtp:97 surround-params="sv << session.surround_params << std::endl;
+    }
+
     for (int x = 0; x < audio::MAX_STREAM_CONFIG; ++x) {
       auto &stream_config = audio::stream_configs[x];
       std::uint8_t mapping[platf::speaker::MAX_SPEAKERS];
@@ -1033,6 +1039,29 @@ namespace rtsp_stream {
         }
       }
     }
+    else if (session.surround_params.length() > 3) {
+      // Channels
+      std::uint8_t c = session.surround_params[0] - '0';
+      // Streams
+      std::uint8_t n = session.surround_params[1] - '0';
+      // Coupled streams
+      std::uint8_t m = session.surround_params[2] - '0';
+      auto valid = false;
+      if ((c == 6 || c == 8) && c == config.audio.channels && n + m == c && session.surround_params.length() == c + 3) {
+        config.audio.customStreamParams.channelCount = c;
+        config.audio.customStreamParams.streams = n;
+        config.audio.customStreamParams.coupledStreams = m;
+        valid = true;
+        for (std::uint8_t i = 0; i < c; i++) {
+          config.audio.customStreamParams.mapping[i] = session.surround_params[i + 3] - '0';
+          if (config.audio.customStreamParams.mapping[i] >= c) {
+            valid = false;
+            break;
+          }
+        }
+      }
+      config.audio.flags[audio::config_t::CUSTOM_SURROUND_PARAMS] = valid;
+    }
 
     // If the client sent a configured bitrate, we will choose the actual bitrate ourselves
     // by using FEC percentage and audio quality settings. If the calculated bitrate ends up

src/rtsp.h

@@ -29,6 +29,7 @@ namespace rtsp_stream {
     int gcmap;
     int appid;
     int surround_info;
+    std::string surround_params;
     bool enable_hdr;
     bool enable_sops;
 

tests/conftest.cpp

@@ -176,7 +176,7 @@ protected:
     std::cout << "PlatformInitTest:: starting Fixture SetUp" << std::endl;
 
     // initialize the platform
-    auto deinit_guard = platf::init();
+    deinit_guard = platf::init();
     if (!deinit_guard) {
       FAIL() << "Platform failed to initialize";
     }
@@ -187,8 +187,12 @@ protected:
   void
   TearDown() override {
     std::cout << "PlatformInitTest:: starting Fixture TearDown" << std::endl;
+    deinit_guard.reset(nullptr);
     std::cout << "PlatformInitTest:: finished Fixture TearDown" << std::endl;
   }
+
+private:
+  std::unique_ptr<platf::deinit_t> deinit_guard;
 };
 
 class DocsPythonVenvBase: public virtual BaseTest {

tests/unit/test_audio.cpp

@@ -0,0 +1,81 @@
+/**
+ * @file tests/test_audio.cpp
+ * @brief Test src/audio.*.
+ */
+#include <bitset>
+#include <src/audio.h>
+
+#include <tests/conftest.cpp>
+
+using namespace audio;
+
+class AudioTest: public virtual BaseTest, public PlatformInitBase, public ::testing::WithParamInterface<std::tuple<std::basic_string_view<char>, config_t>> {
+protected:
+  void
+  SetUp() override {
+    BaseTest::SetUp();
+    PlatformInitBase::SetUp();
+
+    std::string_view p_name = std::get<0>(GetParam());
+    std::cout << "AudioTest(" << p_name << "):: starting Fixture SetUp" << std::endl;
+
+    m_config = std::get<1>(GetParam());
+    m_mail = std::make_shared<safe::mail_raw_t>();
+  }
+
+  void
+  TearDown() override {
+    PlatformInitBase::TearDown();
+    BaseTest::TearDown();
+  }
+
+protected:
+  config_t m_config;
+  safe::mail_t m_mail;
+};
+
+static std::bitset<config_t::MAX_FLAGS>
+config_flags(int flag = -1) {
+  std::bitset<3> result = std::bitset<config_t::MAX_FLAGS>();
+  if (flag >= 0) {
+    result.set(flag);
+  }
+  return result;
+}
+
+INSTANTIATE_TEST_SUITE_P(
+  Configurations,
+  AudioTest,
+  ::testing::Values(
+    std::make_tuple("HIGH_STEREO", config_t { 5, 2, 0x3, { 0 }, config_flags(config_t::HIGH_QUALITY) }),
+    std::make_tuple("SURROUND51", config_t { 5, 6, 0x3F, { 0 }, config_flags() }),
+    std::make_tuple("SURROUND71", config_t { 5, 8, 0x63F, { 0 }, config_flags() }),
+    std::make_tuple("SURROUND51_CUSTOM", config_t { 5, 6, 0x3F, { 6, 4, 2, { 0, 1, 4, 5, 2, 3 } }, config_flags(config_t::CUSTOM_SURROUND_PARAMS) })));
+
+TEST_P(AudioTest, TestEncode) {
+  std::thread timer([&] {
+    // Terminate the audio capture after 5 seconds.
+    std::this_thread::sleep_for(5s);
+    auto shutdown_event = m_mail->event<bool>(mail::shutdown);
+    auto audio_packets = m_mail->queue<packet_t>(mail::audio_packets);
+    shutdown_event->raise(true);
+    audio_packets->stop();
+  });
+  std::thread capture([&] {
+    auto packets = m_mail->queue<packet_t>(mail::audio_packets);
+    auto shutdown_event = m_mail->event<bool>(mail::shutdown);
+    while (auto packet = packets->pop()) {
+      if (shutdown_event->peek()) {
+        break;
+      }
+      auto packet_data = packet->second;
+      if (packet_data.size() == 0) {
+        FAIL() << "Empty packet data";
+      }
+    }
+  });
+  audio::capture(m_mail, m_config, nullptr);
+
+  timer.join();
+  capture.join();
+}