Flesh out audio state machine for playback

Also fix mono playback

src/audio/audio_decoder.cpp

@@ -98,27 +98,27 @@ auto AudioDecoder::Process(const std::vector<InputStream>& inputs,
 
   while (true) {
     if (has_samples_to_send_) {
-      if (!current_output_format_) {
-        auto format = current_codec_->GetOutputFormat();
+      auto format = current_codec_->GetOutputFormat();
+      if (format.has_value()) {
         current_output_format_ = StreamInfo::Pcm{
-            .channels = format.num_channels,
-            .bits_per_sample = format.bits_per_sample,
-            .sample_rate = format.sample_rate_hz,
+            .channels = format->num_channels,
+            .bits_per_sample = format->bits_per_sample,
+            .sample_rate = format->sample_rate_hz,
         };
-      }
 
-      if (!output->prepare(*current_output_format_)) {
-        break;
-      }
+        if (!output->prepare(*current_output_format_)) {
+          break;
+        }
 
-      auto write_res = current_codec_->WriteOutputSamples(output->data());
-      output->add(write_res.first);
-      has_samples_to_send_ = !write_res.second;
+        auto write_res = current_codec_->WriteOutputSamples(output->data());
+        output->add(write_res.first);
+        has_samples_to_send_ = !write_res.second;
 
-      if (has_samples_to_send_) {
-        // We weren't able to fit all the generated samples into the output
-        // buffer. Stop trying; we'll finish up during the next pass.
-        break;
+        if (has_samples_to_send_) {
+          // We weren't able to fit all the generated samples into the output
+          // buffer. Stop trying; we'll finish up during the next pass.
+          break;
+        }
       }
     }
 

src/audio/audio_fsm.cpp

@@ -6,6 +6,7 @@
 
 #include "audio_fsm.hpp"
 #include <memory>
+#include <variant>
 #include "audio_decoder.hpp"
 #include "audio_events.hpp"
 #include "audio_task.hpp"
@@ -16,6 +17,8 @@
 
 namespace audio {
 
+static const char kTag[] = "audio_fsm";
+
 drivers::GpioExpander* AudioState::sGpioExpander;
 std::shared_ptr<drivers::I2SDac> AudioState::sDac;
 std::shared_ptr<drivers::DigitalPot> AudioState::sPots;
@@ -25,6 +28,8 @@ std::unique_ptr<FatfsAudioInput> AudioState::sFileSource;
 std::unique_ptr<I2SAudioOutput> AudioState::sI2SOutput;
 std::vector<std::unique_ptr<IAudioElement>> AudioState::sPipeline;
 
+std::deque<AudioState::EnqueuedItem> AudioState::sSongQueue;
+
 auto AudioState::Init(drivers::GpioExpander* gpio_expander,
                       std::weak_ptr<database::Database> database) -> bool {
   sGpioExpander = gpio_expander;
@@ -66,6 +71,33 @@ void Standby::react(const PlayFile& ev) {
   }
 }
 
+void Playback::entry() {
+  ESP_LOGI(kTag, "beginning playback");
+  sI2SOutput->SetInUse(true);
+}
+
+void Playback::exit() {
+  ESP_LOGI(kTag, "finishing playback");
+  sI2SOutput->SetInUse(false);
+}
+
+void Playback::react(const InputFileFinished& ev) {
+  ESP_LOGI(kTag, "finished file");
+  if (sSongQueue.empty()) {
+    return;
+  }
+  EnqueuedItem next_item = sSongQueue.front();
+  sSongQueue.pop_front();
+
+  if (std::holds_alternative<std::string>(next_item)) {
+    sFileSource->OpenFile(std::get<std::string>(next_item));
+  }
+}
+
+void Playback::react(const AudioPipelineIdle& ev) {
+  transit<Standby>();
+}
+
 }  // namespace states
 
 }  // namespace audio

src/audio/audio_task.cpp

@@ -15,6 +15,8 @@
 #include <memory>
 #include <variant>
 
+#include "audio_events.hpp"
+#include "audio_fsm.hpp"
 #include "audio_sink.hpp"
 #include "cbor.h"
 #include "esp_err.h"
@@ -60,6 +62,7 @@ void AudioTaskMain(std::unique_ptr<Pipeline> pipeline, IAudioSink* sink) {
 
   std::vector<Pipeline*> all_elements = pipeline->GetIterationOrder();
 
+  bool previously_had_work = false;
   events::EventQueue& event_queue = events::EventQueue::GetInstance();
   while (1) {
     // First, see if we actually have any pipeline work to do in this iteration.
@@ -75,6 +78,11 @@ void AudioTaskMain(std::unique_ptr<Pipeline> pipeline, IAudioSink* sink) {
       }
     }
 
+    if (previously_had_work && !has_work) {
+      events::Dispatch<AudioPipelineIdle, AudioState>({});
+    }
+    previously_had_work = has_work;
+
     // See if there's any new events.
     event_queue.ServiceAudio(has_work ? delay_ticks : portMAX_DELAY);
 
@@ -118,6 +126,7 @@ void AudioTaskMain(std::unique_ptr<Pipeline> pipeline, IAudioSink* sink) {
 
     if (sink_stream.info().bytes_in_stream == 0) {
       // No new bytes to sink, so skip sinking completely.
+      ESP_LOGI(kTag, "no bytes to sink");
       continue;
     }
 
@@ -130,6 +139,7 @@ void AudioTaskMain(std::unique_ptr<Pipeline> pipeline, IAudioSink* sink) {
         output_format = sink_stream.info().format;
         sink->Configure(*output_format);
       } else {
+        ESP_LOGI(kTag, "waiting to reconfigure");
         continue;
       }
     }

src/audio/fatfs_audio_input.cpp

@@ -13,8 +13,11 @@
 #include <variant>
 
 #include "arena.hpp"
+#include "audio_events.hpp"
+#include "audio_fsm.hpp"
 #include "esp_heap_caps.h"
 #include "esp_log.h"
+#include "event_queue.hpp"
 #include "ff.h"
 #include "freertos/portmacro.h"
 
@@ -69,6 +72,10 @@ auto FatfsAudioInput::Process(const std::vector<InputStream>& inputs,
   }
 
   std::size_t max_size = output->data().size_bytes();
+  if (max_size < output->data().size_bytes() / 2) {
+    return;
+  }
+
   std::size_t size = 0;
   FRESULT result =
       f_read(&current_file_, output->data().data(), max_size, &size);
@@ -83,6 +90,12 @@ auto FatfsAudioInput::Process(const std::vector<InputStream>& inputs,
   if (size < max_size || f_eof(&current_file_)) {
     f_close(&current_file_);
     is_file_open_ = false;
+
+    // TODO(jacqueline): MP3 only
+    std::fill_n(output->data().begin(), 8, std::byte(0));
+    output->add(8);
+
+    events::Dispatch<InputFileFinished, AudioState>({});
   }
 }
 

src/audio/i2s_audio_output.cpp

@@ -38,7 +38,6 @@ I2SAudioOutput::I2SAudioOutput(drivers::GpioExpander* expander,
       attenuation_(pots_->GetMaxAttenuation()) {
   SetVolume(25);  // For testing
   dac_->SetSource(buffer());
-  dac_->Start();
 }
 
 I2SAudioOutput::~I2SAudioOutput() {
@@ -46,6 +45,15 @@ I2SAudioOutput::~I2SAudioOutput() {
   dac_->SetSource(nullptr);
 }
 
+auto I2SAudioOutput::SetInUse(bool in_use) -> void {
+  if (in_use) {
+    dac_->Start();
+  } else {
+    dac_->Stop();
+  }
+  pots_->SetZeroCrossDetect(in_use);
+}
+
 auto I2SAudioOutput::SetVolumeImbalance(int_fast8_t balance) -> void {
   int_fast8_t new_difference = balance - left_difference_;
   left_difference_ = balance;
@@ -124,6 +132,19 @@ auto I2SAudioOutput::Configure(const StreamInfo::Format& format) -> bool {
   ESP_LOGI(kTag, "incoming audio stream: %u bpp @ %lu Hz", pcm.bits_per_sample,
            pcm.sample_rate);
 
+  drivers::I2SDac::Channels ch;
+  switch (pcm.channels) {
+    case 1:
+      ch = drivers::I2SDac::CHANNELS_MONO;
+      break;
+    case 2:
+      ch = drivers::I2SDac::CHANNELS_STEREO;
+      break;
+    default:
+      ESP_LOGE(kTag, "dropping stream with out of bounds channels");
+      return false;
+  }
+
   drivers::I2SDac::BitsPerSample bps;
   switch (pcm.bits_per_sample) {
     case 16:
@@ -153,9 +174,7 @@ auto I2SAudioOutput::Configure(const StreamInfo::Format& format) -> bool {
       return false;
   }
 
-  // TODO(jacqueline): probs do something with the channel hey
-
-  dac_->Reconfigure(bps, sample_rate);
+  dac_->Reconfigure(ch, bps, sample_rate);
   current_config_ = pcm;
 
   return true;

src/audio/include/audio_events.hpp

@@ -24,4 +24,7 @@ struct PlaySong : tinyfsm::Event {
   std::optional<database::SongTags> tags;
 };
 
+struct InputFileFinished : tinyfsm::Event {};
+struct AudioPipelineIdle : tinyfsm::Event {};
+
 }  // namespace audio

src/audio/include/audio_fsm.hpp

@@ -6,6 +6,7 @@
 
 #pragma once
 
+#include <deque>
 #include <memory>
 #include <vector>
 
@@ -16,6 +17,7 @@
 #include "gpio_expander.hpp"
 #include "i2s_audio_output.hpp"
 #include "i2s_dac.hpp"
+#include "song.hpp"
 #include "storage.hpp"
 #include "tinyfsm.hpp"
 
@@ -40,6 +42,9 @@ class AudioState : public tinyfsm::Fsm<AudioState> {
   virtual void react(const PlaySong&) {}
   virtual void react(const PlayFile&) {}
 
+  virtual void react(const InputFileFinished&) {}
+  virtual void react(const AudioPipelineIdle&) {}
+
  protected:
   static drivers::GpioExpander* sGpioExpander;
   static std::shared_ptr<drivers::I2SDac> sDac;
@@ -49,6 +54,9 @@ class AudioState : public tinyfsm::Fsm<AudioState> {
   static std::unique_ptr<FatfsAudioInput> sFileSource;
   static std::unique_ptr<I2SAudioOutput> sI2SOutput;
   static std::vector<std::unique_ptr<IAudioElement>> sPipeline;
+
+  typedef std::variant<database::SongId, std::string> EnqueuedItem;
+  static std::deque<EnqueuedItem> sSongQueue;
 };
 
 namespace states {
@@ -68,6 +76,12 @@ class Standby : public AudioState {
 
 class Playback : public AudioState {
  public:
+  void entry() override;
+  void exit() override;
+
+  void react(const InputFileFinished&) override;
+  void react(const AudioPipelineIdle&) override;
+
   using AudioState::react;
 };
 

src/audio/include/audio_sink.hpp

@@ -40,6 +40,8 @@ class IAudioSink {
     free(metadata_);
   }
 
+  virtual auto SetInUse(bool) -> void {}
+
   virtual auto SetVolumeImbalance(int_fast8_t balance) -> void = 0;
   virtual auto SetVolume(uint_fast8_t percent) -> void = 0;
   virtual auto GetVolume() -> uint_fast8_t = 0;

src/audio/include/i2s_audio_output.hpp

@@ -29,6 +29,8 @@ class I2SAudioOutput : public IAudioSink {
                  std::weak_ptr<drivers::DigitalPot> pots);
   ~I2SAudioOutput();
 
+  auto SetInUse(bool) -> void override;
+
   auto SetVolumeImbalance(int_fast8_t balance) -> void override;
   auto SetVolume(uint_fast8_t percent) -> void override;
   auto GetVolume() -> uint_fast8_t override;

src/audio/include/pipeline.hpp

@@ -36,6 +36,10 @@ class Pipeline {
 
   auto GetIterationOrder() -> std::vector<Pipeline*>;
 
+  // Not copyable or movable.
+  Pipeline(const Pipeline&) = delete;
+  Pipeline& operator=(const Pipeline&) = delete;
+
  private:
   IAudioElement* root_;
   std::vector<std::unique_ptr<Pipeline>> subtrees_;

src/audio/stream_info.cpp

@@ -22,7 +22,8 @@ namespace audio {
 
 void InputStream::consume(std::size_t bytes) const {
   assert(raw_->info->bytes_in_stream >= bytes);
-  auto new_data = raw_->data.subspan(bytes);
+  auto new_data =
+      raw_->data.subspan(bytes, raw_->info->bytes_in_stream - bytes);
   std::move(new_data.begin(), new_data.end(), raw_->data.begin());
   raw_->info->bytes_in_stream = new_data.size_bytes();
 }

src/codecs/include/codec.hpp

@@ -11,6 +11,7 @@
 #include <cstddef>
 #include <cstdint>
 #include <memory>
+#include <optional>
 #include <string>
 #include <utility>
 
@@ -32,7 +33,7 @@ class ICodec {
     uint32_t sample_rate_hz;
   };
 
-  virtual auto GetOutputFormat() -> OutputFormat = 0;
+  virtual auto GetOutputFormat() -> std::optional<OutputFormat> = 0;
 
   enum ProcessingError { MALFORMED_DATA };
 

src/codecs/include/mad.hpp

@@ -8,6 +8,7 @@
 
 #include <cstddef>
 #include <cstdint>
+#include <optional>
 #include <string>
 #include <utility>
 
@@ -24,7 +25,7 @@ class MadMp3Decoder : public ICodec {
   ~MadMp3Decoder();
 
   auto CanHandleType(StreamType type) -> bool override;
-  auto GetOutputFormat() -> OutputFormat override;
+  auto GetOutputFormat() -> std::optional<OutputFormat> override;
   auto ResetForNewStream() -> void override;
   auto SetInput(cpp::span<const std::byte> input) -> void override;
   auto GetInputPosition() -> std::size_t override;
@@ -37,9 +38,6 @@ class MadMp3Decoder : public ICodec {
   mad_frame frame_;
   mad_synth synth_;
 
-  mad_header header_;
-  bool has_decoded_header_;
-
   int current_sample_;
 };
 

src/codecs/mad.cpp

@@ -8,6 +8,7 @@
 #include <stdint.h>
 
 #include <cstdint>
+#include <optional>
 
 #include "mad.h"
 
@@ -34,31 +35,29 @@ MadMp3Decoder::MadMp3Decoder() {
   mad_stream_init(&stream_);
   mad_frame_init(&frame_);
   mad_synth_init(&synth_);
-  mad_header_init(&header_);
 }
 MadMp3Decoder::~MadMp3Decoder() {
   mad_stream_finish(&stream_);
   mad_frame_finish(&frame_);
   mad_synth_finish(&synth_);
-  mad_header_finish(&header_);
 }
 
 auto MadMp3Decoder::CanHandleType(StreamType type) -> bool {
   return type == STREAM_MP3;
 }
 
-auto MadMp3Decoder::GetOutputFormat() -> OutputFormat {
-  return OutputFormat{
+auto MadMp3Decoder::GetOutputFormat() -> std::optional<OutputFormat> {
+  if (synth_.pcm.channels == 0 || synth_.pcm.samplerate == 0) {
+    return {};
+  }
+  return std::optional<OutputFormat>({
       .num_channels = static_cast<uint8_t>(synth_.pcm.channels),
-      .bits_per_sample = 16,
-      .sample_rate_hz =
-          synth_.pcm.samplerate == 0 ? 44100 : synth_.pcm.samplerate,
-  };
+      .bits_per_sample = 24,
+      .sample_rate_hz = synth_.pcm.samplerate,
+  });
 }
 
-auto MadMp3Decoder::ResetForNewStream() -> void {
-  has_decoded_header_ = false;
-}
+auto MadMp3Decoder::ResetForNewStream() -> void {}
 
 auto MadMp3Decoder::SetInput(cpp::span<const std::byte> input) -> void {
   mad_stream_buffer(&stream_,
@@ -71,16 +70,6 @@ auto MadMp3Decoder::GetInputPosition() -> std::size_t {
 }
 
 auto MadMp3Decoder::ProcessNextFrame() -> cpp::result<bool, ProcessingError> {
-  if (!has_decoded_header_) {
-    // The header of any given frame should be representative of the
-    // entire stream, so only need to read it once.
-    mad_header_decode(&header_, &stream_);
-    has_decoded_header_ = true;
-
-    // TODO: Use the info in the header for something. I think the
-    // duration will help with seeking?
-  }
-
   // Whatever was last synthesized is now invalid, so ensure we don't try to
   // send it.
   current_sample_ = -1;
@@ -128,7 +117,6 @@ auto MadMp3Decoder::WriteOutputSamples(cpp::span<std::byte> output)
     for (int channel = 0; channel < synth_.pcm.channels; channel++) {
       // TODO(jacqueline): output 24 bit samples when (if?) we have a downmix
       // step in the pipeline.
-      /*
       uint32_t sample_24 =
           scaleToBits(synth_.pcm.samples[channel][current_sample_], 24);
       output[output_byte++] = static_cast<std::byte>((sample_24 >> 16) & 0xFF);
@@ -136,11 +124,12 @@ auto MadMp3Decoder::WriteOutputSamples(cpp::span<std::byte> output)
       output[output_byte++] = static_cast<std::byte>((sample_24)&0xFF);
       // 24 bit samples must still be aligned to 32 bits. The LSB is ignored.
       output[output_byte++] = static_cast<std::byte>(0);
-      */
+      /*
       uint16_t sample_16 =
           scaleToBits(synth_.pcm.samples[channel][current_sample_], 16);
       output[output_byte++] = static_cast<std::byte>((sample_16 >> 8) & 0xFF);
       output[output_byte++] = static_cast<std::byte>((sample_16)&0xFF);
+      */
     }
     current_sample_++;
   }

src/drivers/display.cpp

@@ -126,7 +126,7 @@ auto Display::Create(GpioExpander* expander,
   ESP_ERROR_CHECK(ledc_set_duty(LEDC_LOW_SPEED_MODE, LEDC_CHANNEL_0, 0));
   ESP_ERROR_CHECK(ledc_update_duty(LEDC_LOW_SPEED_MODE, LEDC_CHANNEL_0));
 
-  ledc_fade_func_install(0);
+  // ledc_fade_func_install(0);
 
   // Next, init the SPI device
   spi_device_interface_config_t spi_cfg = {
@@ -194,8 +194,9 @@ auto Display::SetDisplayOn(bool enabled) -> void {
   display_on_ = enabled;
 
   int new_duty = display_on_ ? brightness_ : 0;
-  ledc_set_fade_with_time(LEDC_LOW_SPEED_MODE, LEDC_CHANNEL_0, new_duty, 250);
-  ledc_fade_start(LEDC_LOW_SPEED_MODE, LEDC_CHANNEL_0, LEDC_FADE_NO_WAIT);
+  // ledc_set_fade_with_time(LEDC_LOW_SPEED_MODE, LEDC_CHANNEL_0, new_duty,
+  // 250); ledc_fade_start(LEDC_LOW_SPEED_MODE, LEDC_CHANNEL_0,
+  // LEDC_FADE_NO_WAIT);
 }
 
 void Display::SendInitialisationSequence(const uint8_t* data) {

src/drivers/i2s_dac.cpp

@@ -116,24 +116,30 @@ auto I2SDac::Stop() -> void {
   i2s_active_ = false;
 }
 
-auto I2SDac::Reconfigure(BitsPerSample bps, SampleRate rate) -> void {
+auto I2SDac::Reconfigure(Channels ch, BitsPerSample bps, SampleRate rate)
+    -> void {
   if (i2s_active_) {
     i2s_channel_disable(i2s_handle_);
   }
 
-  uint8_t bps_bits = 0;
+  switch (ch) {
+    case CHANNELS_MONO:
+      slot_config_.slot_mode = I2S_SLOT_MODE_MONO;
+      break;
+    case CHANNELS_STEREO:
+      slot_config_.slot_mode = I2S_SLOT_MODE_STEREO;
+      break;
+  }
+
   switch (bps) {
     case BPS_16:
       slot_config_.data_bit_width = I2S_DATA_BIT_WIDTH_16BIT;
-      bps_bits = 0;
       break;
     case BPS_24:
       slot_config_.data_bit_width = I2S_DATA_BIT_WIDTH_24BIT;
-      bps_bits = 0b10;
       break;
     case BPS_32:
       slot_config_.data_bit_width = I2S_DATA_BIT_WIDTH_32BIT;
-      bps_bits = 0b11;
       break;
   }
   ESP_ERROR_CHECK(i2s_channel_reconfig_std_slot(i2s_handle_, &slot_config_));

src/drivers/include/display.hpp

@@ -43,6 +43,10 @@ class Display {
                    const lv_area_t* area,
                    lv_color_t* color_map);
 
+  // Not copyable or movable.
+  Display(const Display&) = delete;
+  Display& operator=(const Display&) = delete;
+
  private:
   GpioExpander* gpio_;
   spi_device_handle_t handle_;

src/drivers/include/i2s_dac.hpp

@@ -40,6 +40,10 @@ class I2SDac {
   auto Start() -> void;
   auto Stop() -> void;
 
+  enum Channels {
+    CHANNELS_MONO,
+    CHANNELS_STEREO,
+  };
   enum BitsPerSample {
     BPS_16 = I2S_DATA_BIT_WIDTH_16BIT,
     BPS_24 = I2S_DATA_BIT_WIDTH_24BIT,
@@ -56,7 +60,7 @@ class I2SDac {
     SAMPLE_RATE_192 = 192000,
   };
 
-  auto Reconfigure(BitsPerSample bps, SampleRate rate) -> void;
+  auto Reconfigure(Channels ch, BitsPerSample bps, SampleRate rate) -> void;
 
   auto WriteData(const cpp::span<const std::byte>& data) -> void;
   auto SetSource(StreamBufferHandle_t buffer) -> void;