diff --git a/src/audio/CMakeLists.txt b/src/audio/CMakeLists.txt
index de6c9b64..df5622f5 100644
--- a/src/audio/CMakeLists.txt
+++ b/src/audio/CMakeLists.txt
@@ -3,8 +3,8 @@
 # SPDX-License-Identifier: GPL-3.0-only
 
 idf_component_register(
-  SRCS "audio_task.cpp" "fatfs_audio_input.cpp" "i2s_audio_output.cpp"
-  "track_queue.cpp" "audio_fsm.cpp" "sink_mixer.cpp" "resample.cpp"
+  SRCS "audio_decoder.cpp" "fatfs_audio_input.cpp" "i2s_audio_output.cpp"
+  "track_queue.cpp" "audio_fsm.cpp" "audio_converter.cpp" "resample.cpp"
   "fatfs_source.cpp" "bt_audio_output.cpp"
   INCLUDE_DIRS "include"
   REQUIRES "codecs" "drivers" "cbor" "result" "tasks" "span" "memory" "tinyfsm"
diff --git a/src/audio/sink_mixer.cpp b/src/audio/audio_converter.cpp
similarity index 92%
rename from src/audio/sink_mixer.cpp
rename to src/audio/audio_converter.cpp
index ad7198dc..c540d821 100644
--- a/src/audio/sink_mixer.cpp
+++ b/src/audio/audio_converter.cpp
@@ -4,7 +4,7 @@
  * SPDX-License-Identifier: GPL-3.0-only
  */
 
-#include "sink_mixer.hpp"
+#include "audio_converter.hpp"
 
 #include <algorithm>
 #include <cmath>
@@ -28,7 +28,7 @@ static constexpr std::size_t kSampleBufferLength = 240 * 2;
 
 namespace audio {
 
-SinkMixer::SinkMixer()
+SampleConverter::SampleConverter()
     : commands_(xQueueCreate(1, sizeof(Args))),
       resampler_(nullptr),
       source_(xStreamBufferCreateWithCaps(kSourceBufferLength,
@@ -49,21 +49,21 @@ SinkMixer::SinkMixer()
   tasks::StartPersistent<tasks::Type::kMixer>([&]() { Main(); });
 }
 
-SinkMixer::~SinkMixer() {
+SampleConverter::~SampleConverter() {
   vQueueDelete(commands_);
   vStreamBufferDelete(source_);
 }
 
-auto SinkMixer::SetOutput(std::shared_ptr<IAudioOutput> output) -> void {
+auto SampleConverter::SetOutput(std::shared_ptr<IAudioOutput> output) -> void {
   // FIXME: We should add synchronisation here, but we should be careful about
   // not impacting performance given that the output will change only very
   // rarely (if ever).
   sink_ = output;
 }
 
-auto SinkMixer::MixAndSend(cpp::span<sample::Sample> input,
-                           const IAudioOutput::Format& format,
-                           bool is_eos) -> void {
+auto SampleConverter::ConvertSamples(cpp::span<sample::Sample> input,
+                                     const IAudioOutput::Format& format,
+                                     bool is_eos) -> void {
   Args args{
       .format = format,
       .samples_available = input.size(),
@@ -81,7 +81,7 @@ auto SinkMixer::MixAndSend(cpp::span<sample::Sample> input,
   }
 }
 
-auto SinkMixer::Main() -> void {
+auto SampleConverter::Main() -> void {
   for (;;) {
     Args args;
     while (!xQueueReceive(commands_, &args, portMAX_DELAY)) {
@@ -149,8 +149,8 @@ auto SinkMixer::Main() -> void {
   }
 }
 
-auto SinkMixer::HandleSamples(cpp::span<sample::Sample> input, bool is_eos)
-    -> size_t {
+auto SampleConverter::HandleSamples(cpp::span<sample::Sample> input,
+                                    bool is_eos) -> size_t {
   if (source_format_ == target_format_) {
     // The happiest possible case: the input format matches the output
     // format already.
diff --git a/src/audio/audio_task.cpp b/src/audio/audio_decoder.cpp
similarity index 81%
rename from src/audio/audio_task.cpp
rename to src/audio/audio_decoder.cpp
index 99b1c170..03f81124 100644
--- a/src/audio/audio_task.cpp
+++ b/src/audio/audio_decoder.cpp
@@ -4,7 +4,7 @@
  * SPDX-License-Identifier: GPL-3.0-only
  */
 
-#include "audio_task.hpp"
+#include "audio_decoder.hpp"
 
 #include <cstdint>
 #include <cstdlib>
@@ -28,6 +28,7 @@
 #include "freertos/ringbuf.h"
 #include "span.hpp"
 
+#include "audio_converter.hpp"
 #include "audio_events.hpp"
 #include "audio_fsm.hpp"
 #include "audio_sink.hpp"
@@ -36,7 +37,6 @@
 #include "event_queue.hpp"
 #include "fatfs_audio_input.hpp"
 #include "sample.hpp"
-#include "sink_mixer.hpp"
 #include "tasks.hpp"
 #include "track.hpp"
 #include "types.hpp"
@@ -76,23 +76,27 @@ auto Timer::AddSamples(std::size_t samples) -> void {
   }
 }
 
-auto AudioTask::Start(std::shared_ptr<IAudioSource> source,
-                      std::shared_ptr<SinkMixer> sink) -> AudioTask* {
-  AudioTask* task = new AudioTask(source, sink);
+auto Decoder::Start(std::shared_ptr<IAudioSource> source,
+                    std::shared_ptr<SampleConverter> sink) -> Decoder* {
+  Decoder* task = new Decoder(source, sink);
   tasks::StartPersistent<tasks::Type::kAudio>([=]() { task->Main(); });
   return task;
 }
 
-AudioTask::AudioTask(std::shared_ptr<IAudioSource> source,
-                     std::shared_ptr<SinkMixer> mixer)
-    : source_(source), mixer_(mixer), codec_(), timer_(), current_format_() {
+Decoder::Decoder(std::shared_ptr<IAudioSource> source,
+                 std::shared_ptr<SampleConverter> mixer)
+    : source_(source),
+      converter_(mixer),
+      codec_(),
+      timer_(),
+      current_format_() {
   codec_buffer_ = {
       reinterpret_cast<sample::Sample*>(heap_caps_calloc(
           kCodecBufferLength, sizeof(sample::Sample), MALLOC_CAP_SPIRAM)),
       kCodecBufferLength};
 }
 
-void AudioTask::Main() {
+void Decoder::Main() {
   for (;;) {
     if (source_->HasNewStream() || !stream_) {
       std::shared_ptr<codecs::IStream> new_stream = source_->NextStream();
@@ -110,7 +114,7 @@ void AudioTask::Main() {
   }
 }
 
-auto AudioTask::BeginDecoding(std::shared_ptr<codecs::IStream> stream) -> bool {
+auto Decoder::BeginDecoding(std::shared_ptr<codecs::IStream> stream) -> bool {
   codec_.reset(codecs::CreateCodecForType(stream->type()).value_or(nullptr));
   if (!codec_) {
     ESP_LOGE(kTag, "no codec found");
@@ -140,15 +144,16 @@ auto AudioTask::BeginDecoding(std::shared_ptr<codecs::IStream> stream) -> bool {
   return true;
 }
 
-auto AudioTask::ContinueDecoding() -> bool {
+auto Decoder::ContinueDecoding() -> bool {
   auto res = codec_->DecodeTo(codec_buffer_);
   if (res.has_error()) {
     return true;
   }
 
   if (res->samples_written > 0) {
-    mixer_->MixAndSend(codec_buffer_.first(res->samples_written),
-                       current_sink_format_.value(), res->is_stream_finished);
+    converter_->ConvertSamples(codec_buffer_.first(res->samples_written),
+                               current_sink_format_.value(),
+                               res->is_stream_finished);
   }
 
   if (timer_) {
diff --git a/src/audio/audio_fsm.cpp b/src/audio/audio_fsm.cpp
index 1ea670af..e68eedaf 100644
--- a/src/audio/audio_fsm.cpp
+++ b/src/audio/audio_fsm.cpp
@@ -15,8 +15,9 @@
 #include "freertos/portmacro.h"
 #include "freertos/projdefs.h"
 
+#include "audio_converter.hpp"
+#include "audio_decoder.hpp"
 #include "audio_events.hpp"
-#include "audio_task.hpp"
 #include "bluetooth.hpp"
 #include "bt_audio_output.hpp"
 #include "event_queue.hpp"
@@ -24,7 +25,6 @@
 #include "future_fetcher.hpp"
 #include "i2s_audio_output.hpp"
 #include "i2s_dac.hpp"
-#include "sink_mixer.hpp"
 #include "system_events.hpp"
 #include "track.hpp"
 #include "track_queue.hpp"
@@ -37,10 +37,9 @@ drivers::IGpios* AudioState::sIGpios;
 std::shared_ptr<drivers::I2SDac> AudioState::sDac;
 std::weak_ptr<database::Database> AudioState::sDatabase;
 
-std::unique_ptr<AudioTask> AudioState::sTask;
-
 std::shared_ptr<FatfsAudioInput> AudioState::sFileSource;
-std::shared_ptr<SinkMixer> AudioState::sMixer;
+std::unique_ptr<Decoder> AudioState::sDecoder;
+std::shared_ptr<SampleConverter> AudioState::sSampleConverter;
 std::shared_ptr<IAudioOutput> AudioState::sOutput;
 
 TrackQueue* AudioState::sTrackQueue;
@@ -65,10 +64,10 @@ auto AudioState::Init(drivers::IGpios* gpio_expander,
   sOutput.reset(new I2SAudioOutput(sIGpios, sDac));
   // sOutput.reset(new BluetoothAudioOutput(bluetooth));
 
-  sMixer.reset(new SinkMixer());
-  sMixer->SetOutput(sOutput);
+  sSampleConverter.reset(new SampleConverter());
+  sSampleConverter->SetOutput(sOutput);
 
-  AudioTask::Start(sFileSource, sMixer);
+  Decoder::Start(sFileSource, sSampleConverter);
 
   return true;
 }
diff --git a/src/audio/include/sink_mixer.hpp b/src/audio/include/audio_converter.hpp
similarity index 69%
rename from src/audio/include/sink_mixer.hpp
rename to src/audio/include/audio_converter.hpp
index d046f835..81532969 100644
--- a/src/audio/include/sink_mixer.hpp
+++ b/src/audio/include/audio_converter.hpp
@@ -18,19 +18,21 @@
 namespace audio {
 
 /*
- * Handles the final downmix + resample + quantisation stage of audio,
- * generation sending the result directly to an IAudioOutput.
+ * Handle to a persistent task that converts samples between formats (sample
+ * rate, channels, bits per sample), in order to put samples in the preferred
+ * format of the current output device. The resulting samples are forwarded
+ * to the output device's sink stream.
  */
-class SinkMixer {
+class SampleConverter {
  public:
-  SinkMixer();
-  ~SinkMixer();
+  SampleConverter();
+  ~SampleConverter();
 
   auto SetOutput(std::shared_ptr<IAudioOutput>) -> void;
 
-  auto MixAndSend(cpp::span<sample::Sample>,
-                  const IAudioOutput::Format& format,
-                  bool is_eos) -> void;
+  auto ConvertSamples(cpp::span<sample::Sample>,
+                      const IAudioOutput::Format& format,
+                      bool is_eos) -> void;
 
  private:
   auto Main() -> void;
diff --git a/src/audio/include/audio_task.hpp b/src/audio/include/audio_decoder.hpp
similarity index 63%
rename from src/audio/include/audio_task.hpp
rename to src/audio/include/audio_decoder.hpp
index 08c5769c..1759f6e4 100644
--- a/src/audio/include/audio_task.hpp
+++ b/src/audio/include/audio_decoder.hpp
@@ -9,15 +9,18 @@
 #include <cstdint>
 #include <memory>
 
+#include "audio_converter.hpp"
 #include "audio_sink.hpp"
 #include "audio_source.hpp"
 #include "codec.hpp"
-#include "sink_mixer.hpp"
 #include "track.hpp"
 #include "types.hpp"
 
 namespace audio {
 
+/*
+ * Sample-based timer for the current elapsed playback time.
+ */
 class Timer {
  public:
   Timer(const codecs::ICodec::OutputFormat& format);
@@ -32,25 +35,30 @@ class Timer {
   uint32_t total_duration_seconds_;
 };
 
-class AudioTask {
+/*
+ * Handle to a persistent task that takes bytes from the given source, decodes
+ * them into sample::Sample (normalised to 16 bit signed PCM), and then
+ * forwards the resulting stream to the given converter.
+ */
+class Decoder {
  public:
   static auto Start(std::shared_ptr<IAudioSource> source,
-                    std::shared_ptr<SinkMixer> mixer) -> AudioTask*;
+                    std::shared_ptr<SampleConverter> converter) -> Decoder*;
 
   auto Main() -> void;
 
-  AudioTask(const AudioTask&) = delete;
-  AudioTask& operator=(const AudioTask&) = delete;
+  Decoder(const Decoder&) = delete;
+  Decoder& operator=(const Decoder&) = delete;
 
  private:
-  AudioTask(std::shared_ptr<IAudioSource> source,
-            std::shared_ptr<SinkMixer> mixer);
+  Decoder(std::shared_ptr<IAudioSource> source,
+          std::shared_ptr<SampleConverter> converter);
 
   auto BeginDecoding(std::shared_ptr<codecs::IStream>) -> bool;
   auto ContinueDecoding() -> bool;
 
   std::shared_ptr<IAudioSource> source_;
-  std::shared_ptr<SinkMixer> mixer_;
+  std::shared_ptr<SampleConverter> converter_;
 
   std::shared_ptr<codecs::IStream> stream_;
   std::unique_ptr<codecs::ICodec> codec_;
diff --git a/src/audio/include/audio_fsm.hpp b/src/audio/include/audio_fsm.hpp
index 6c785426..430bc298 100644
--- a/src/audio/include/audio_fsm.hpp
+++ b/src/audio/include/audio_fsm.hpp
@@ -13,8 +13,8 @@
 #include "audio_sink.hpp"
 #include "tinyfsm.hpp"
 
+#include "audio_decoder.hpp"
 #include "audio_events.hpp"
-#include "audio_task.hpp"
 #include "bt_audio_output.hpp"
 #include "database.hpp"
 #include "display.hpp"
@@ -68,9 +68,9 @@ class AudioState : public tinyfsm::Fsm<AudioState> {
   static std::shared_ptr<drivers::I2SDac> sDac;
   static std::weak_ptr<database::Database> sDatabase;
 
-  static std::unique_ptr<AudioTask> sTask;
   static std::shared_ptr<FatfsAudioInput> sFileSource;
-  static std::shared_ptr<SinkMixer> sMixer;
+  static std::unique_ptr<Decoder> sDecoder;
+  static std::shared_ptr<SampleConverter> sSampleConverter;
   static std::shared_ptr<IAudioOutput> sOutput;
 
   static TrackQueue* sTrackQueue;