From aaa17f9730c399b39458d03b90824363a97d1072 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jean-Micha=C3=ABl=20Celerier?=
 <jeanmichael.celerier@gmail.com>
Date: Tue, 26 Nov 2024 11:21:58 -0500
Subject: [PATCH] examples: add an audio-to-osc example

---
 cmake/avendish.examples.cmake              |   9 ++
 examples/Advanced/Utilities/AudioToOSC.hpp | 165 +++++++++++++++++++++
 2 files changed, 174 insertions(+)
 create mode 100644 examples/Advanced/Utilities/AudioToOSC.hpp

diff --git a/cmake/avendish.examples.cmake b/cmake/avendish.examples.cmake
index aa5fada..095bcf3 100644
--- a/cmake/avendish.examples.cmake
+++ b/cmake/avendish.examples.cmake
@@ -285,6 +285,15 @@ foreach(theTarget ${OSSIA_EXAMPLES})
   )
 endforeach()
 
+###### Advanced examples
+avnd_make_all(
+  TARGET AudioToOSC
+  MAIN_FILE examples/Advanced/Utilities/AudioToOSC.hpp
+  MAIN_CLASS ad::AudioToOSC
+  C_NAME ad_audio_to_osc
+  LINK_LIBRARIES samplerate ossia::ossia
+)
+
 ###### Ports
 
 ## Essentia
diff --git a/examples/Advanced/Utilities/AudioToOSC.hpp b/examples/Advanced/Utilities/AudioToOSC.hpp
new file mode 100644
index 0000000..680d837
--- /dev/null
+++ b/examples/Advanced/Utilities/AudioToOSC.hpp
@@ -0,0 +1,165 @@
+#pragma once
+#include <fmt/format.h>
+#include <halp/audio.hpp>
+#include <halp/callback.hpp>
+#include <halp/controls.hpp>
+#include <halp/meta.hpp>
+#include <ossia/network/common/complex_type.hpp>
+#include <ossia/network/context.hpp>
+#include <ossia/network/generic/generic_device.hpp>
+#include <ossia/protocols/osc/osc_factory.hpp>
+#include <readerwritercircularbuffer.h>
+#include <samplerate.h>
+
+#include <memory>
+#include <thread>
+
+namespace ad
+{
+struct resampler
+{
+  explicit resampler() noexcept
+      : src{src_new(SRC_SINC_BEST_QUALITY, 1, nullptr)}
+      , data(1024)
+  {
+  }
+  resampler(resampler&&) noexcept = default;
+  resampler& operator=(resampler&&) noexcept = default;
+  resampler(const resampler&) noexcept = delete;
+  resampler& operator=(const resampler&) noexcept = delete;
+
+  ~resampler() { }
+
+  using src_handle
+      = std::unique_ptr<SRC_STATE, decltype([](SRC_STATE* p) { src_delete(p); })>;
+  src_handle src;
+  moodycamel::BlockingReaderWriterCircularBuffer<float> data;
+};
+
+struct osc_thread
+{
+  explicit osc_thread(
+      std::string host, uint16_t port, std::span<std::shared_ptr<resampler>> resamplers)
+      : m_config{.mode = conf::MIRROR, .version = conf::OSC1_1, .transport = ossia::net::udp_configuration{{.local = std::nullopt, .remote = ossia::net::send_socket_configuration{{host, port}}}}}
+      , m_device{ossia::net::make_osc_protocol(m_ctx, m_config), "P"}
+      , m_resamplers{resamplers.begin(), resamplers.end()}
+      , m_sema{0}
+  {
+    auto& root = m_device.get_root_node();
+    for(int i = 0; i < resamplers.size(); i++)
+    {
+      auto channel
+          = ossia::create_parameter(root, fmt::format("/channel.{}", i), "float");
+      m_channels.push_back(channel);
+    }
+
+    m_thread = std::thread([this] {
+      while(!m_stop_flag.test(std::memory_order_acquire))
+      {
+        for(int k = 0; k < m_resamplers.size(); k++)
+        {
+          auto& queue = m_resamplers[k]->data;
+          auto& osc = *m_channels[k];
+          float sample{};
+          if(queue.try_dequeue(sample))
+          {
+            osc.push_value(sample);
+          }
+        }
+      }
+      m_sema.release();
+    });
+  }
+
+  ~osc_thread()
+  {
+    m_stop_flag.test_and_set(std::memory_order_release);
+    m_sema.acquire();
+    m_thread.join();
+  }
+
+  using conf = ossia::net::osc_protocol_configuration;
+  conf m_config;
+
+  std::shared_ptr<ossia::net::network_context> m_ctx
+      = std::make_shared<ossia::net::network_context>();
+  ossia::net::generic_device m_device{};
+  std::vector<ossia::net::parameter_base*> m_channels;
+
+  std::vector<std::shared_ptr<resampler>> m_resamplers;
+
+  std::thread m_thread;
+  std::atomic_flag m_stop_flag = ATOMIC_FLAG_INIT;
+  std::binary_semaphore m_sema;
+};
+
+// The core audio object
+struct AudioToOSC
+{
+public:
+  halp_meta(name, "Audio to OSC")
+  halp_meta(c_name, "audio_to_osc")
+  halp_meta(author, "Jean-Michaël Celerier")
+  halp_meta(category, "Audio/Effects")
+  halp_meta(description, "Convert audio to OSC at a specific rate")
+  halp_meta(uuid, "e525a387-b2ce-43b1-83c6-e308a4d7fd2f")
+
+  struct
+  {
+    halp::dynamic_audio_bus<"In", float> audio;
+    halp::knob_f32<"Frequency", halp::range{1., 1000., 401.2}> frequency;
+  } inputs;
+
+  struct outputs
+  {
+
+  } outputs;
+
+  halp::setup s;
+  void prepare(halp::setup info) noexcept
+  {
+    s = info;
+
+    m_osc.reset();
+    m_resamplers.clear();
+    for(int i = 0; i < info.input_channels; i++)
+      m_resamplers.push_back(std::make_shared<resampler>());
+    m_osc = std::make_shared<osc_thread>("127.0.0.1", 1234, m_resamplers);
+  }
+
+  void operator()(int frames) noexcept
+  {
+    if(s.rate <= 0.)
+      return;
+
+    m_buffer.resize(4 * frames, boost::container::default_init);
+
+    for(std::size_t i = 0; i < s.input_channels; ++i)
+    {
+      SRC_DATA data;
+      auto& src = this->m_resamplers[i]->src;
+      data.data_in = inputs.audio.samples[i];
+      data.data_out = m_buffer.data();
+      data.input_frames = frames;
+      data.output_frames = m_buffer.size();
+      data.input_frames_used = 0;
+      data.output_frames_gen = 0;
+      data.src_ratio = inputs.frequency.value / s.rate;
+      data.end_of_input = 0;
+
+      // Resample
+      src_process(src.get(), &data);
+
+      // Put in ring buffer
+      auto& queue = m_resamplers[i]->data;
+      for(auto sample : std::span(m_buffer.data(), data.output_frames_gen))
+        queue.try_enqueue(sample);
+    }
+  }
+
+private:
+  boost::container::vector<float> m_buffer;
+  std::vector<std::shared_ptr<resampler>> m_resamplers;
+  std::shared_ptr<osc_thread> m_osc;
+};
+}