diff --git a/esphome/components/audio/audio_transfer_buffer.cpp b/esphome/components/audio/audio_transfer_buffer.cpp
new file mode 100644
index 0000000000..9b6067aac4
--- /dev/null
+++ b/esphome/components/audio/audio_transfer_buffer.cpp
@@ -0,0 +1,165 @@
+#include "audio_transfer_buffer.h"
+
+#ifdef USE_ESP32
+
+#include "esphome/core/helpers.h"
+
+namespace esphome {
+namespace audio {
+
+AudioTransferBuffer::~AudioTransferBuffer() { this->deallocate_buffer_(); };
+
+std::unique_ptr<AudioSinkTransferBuffer> AudioSinkTransferBuffer::create(size_t buffer_size) {
+  std::unique_ptr<AudioSinkTransferBuffer> sink_buffer = make_unique<AudioSinkTransferBuffer>();
+
+  if (!sink_buffer->allocate_buffer_(buffer_size)) {
+    return nullptr;
+  }
+
+  return sink_buffer;
+}
+
+std::unique_ptr<AudioSourceTransferBuffer> AudioSourceTransferBuffer::create(size_t buffer_size) {
+  std::unique_ptr<AudioSourceTransferBuffer> source_buffer = make_unique<AudioSourceTransferBuffer>();
+
+  if (!source_buffer->allocate_buffer_(buffer_size)) {
+    return nullptr;
+  }
+
+  return source_buffer;
+}
+
+size_t AudioTransferBuffer::free() const {
+  if (this->buffer_size_ == 0) {
+    return 0;
+  }
+  return this->buffer_size_ - (this->buffer_length_ - (this->data_start_ - this->buffer_));
+}
+
+void AudioTransferBuffer::decrease_buffer_length(size_t bytes) {
+  this->buffer_length_ -= bytes;
+  this->data_start_ += bytes;
+}
+
+void AudioTransferBuffer::increase_buffer_length(size_t bytes) { this->buffer_length_ += bytes; }
+
+void AudioTransferBuffer::clear_buffered_data() {
+  this->buffer_length_ = 0;
+  if (this->ring_buffer_.use_count() > 0) {
+    this->ring_buffer_->reset();
+  }
+}
+
+void AudioSinkTransferBuffer::clear_buffered_data() {
+  this->buffer_length_ = 0;
+  if (this->ring_buffer_.use_count() > 0) {
+    this->ring_buffer_->reset();
+  }
+#ifdef USE_SPEAKER
+  if (this->speaker_ != nullptr) {
+    this->speaker_->stop();
+  }
+#endif
+}
+
+bool AudioTransferBuffer::has_buffered_data() const {
+  if (this->ring_buffer_.use_count() > 0) {
+    return ((this->ring_buffer_->available() > 0) || (this->available() > 0));
+  }
+  return (this->available() > 0);
+}
+
+bool AudioTransferBuffer::reallocate(size_t new_buffer_size) {
+  if (this->buffer_length_ > 0) {
+    // Already has data in the buffer, fail
+    return false;
+  }
+  this->deallocate_buffer_();
+  return this->allocate_buffer_(new_buffer_size);
+}
+
+bool AudioTransferBuffer::allocate_buffer_(size_t buffer_size) {
+  this->buffer_size_ = buffer_size;
+
+  RAMAllocator<uint8_t> allocator(ExternalRAMAllocator<uint8_t>::ALLOW_FAILURE);
+
+  this->buffer_ = allocator.allocate(this->buffer_size_);
+  if (this->buffer_ == nullptr) {
+    return false;
+  }
+
+  this->data_start_ = this->buffer_;
+  this->buffer_length_ = 0;
+
+  return true;
+}
+
+void AudioTransferBuffer::deallocate_buffer_() {
+  if (this->buffer_ != nullptr) {
+    RAMAllocator<uint8_t> allocator(ExternalRAMAllocator<uint8_t>::ALLOW_FAILURE);
+    allocator.deallocate(this->buffer_, this->buffer_size_);
+    this->buffer_ = nullptr;
+    this->data_start_ = nullptr;
+  }
+
+  this->buffer_size_ = 0;
+  this->buffer_length_ = 0;
+}
+
+size_t AudioSourceTransferBuffer::transfer_data_from_source(TickType_t ticks_to_wait) {
+  // Shift data in buffer to start
+  if (this->buffer_length_ > 0) {
+    memmove(this->buffer_, this->data_start_, this->buffer_length_);
+  }
+  this->data_start_ = this->buffer_;
+
+  size_t bytes_to_read = this->free();
+  size_t bytes_read = 0;
+  if (bytes_to_read > 0) {
+    if (this->ring_buffer_.use_count() > 0) {
+      bytes_read = this->ring_buffer_->read((void *) this->get_buffer_end(), bytes_to_read, ticks_to_wait);
+    }
+
+    this->increase_buffer_length(bytes_read);
+  }
+  return bytes_read;
+}
+
+size_t AudioSinkTransferBuffer::transfer_data_to_sink(TickType_t ticks_to_wait) {
+  size_t bytes_written = 0;
+  if (this->available()) {
+#ifdef USE_SPEAKER
+    if (this->speaker_ != nullptr) {
+      bytes_written = this->speaker_->play(this->data_start_, this->available(), ticks_to_wait);
+    } else
+#endif
+        if (this->ring_buffer_.use_count() > 0) {
+      bytes_written =
+          this->ring_buffer_->write_without_replacement((void *) this->data_start_, this->available(), ticks_to_wait);
+    }
+
+    this->decrease_buffer_length(bytes_written);
+
+    // Shift unwritten data to the start of the buffer
+    memmove(this->buffer_, this->data_start_, this->buffer_length_);
+    this->data_start_ = this->buffer_;
+  }
+  return bytes_written;
+}
+
+bool AudioSinkTransferBuffer::has_buffered_data() const {
+#ifdef USE_SPEAKER
+  if (this->speaker_ != nullptr) {
+    return (this->speaker_->has_buffered_data() || (this->available() > 0));
+  }
+#endif
+  if (this->ring_buffer_.use_count() > 0) {
+    return ((this->ring_buffer_->available() > 0) || (this->available() > 0));
+  }
+  return (this->available() > 0);
+}
+
+}  // namespace audio
+}  // namespace esphome
+
+#endif
diff --git a/esphome/components/audio/audio_transfer_buffer.h b/esphome/components/audio/audio_transfer_buffer.h
new file mode 100644
index 0000000000..4e461db56d
--- /dev/null
+++ b/esphome/components/audio/audio_transfer_buffer.h
@@ -0,0 +1,139 @@
+#pragma once
+
+#ifdef USE_ESP32
+#include "esphome/core/defines.h"
+#include "esphome/core/ring_buffer.h"
+
+#ifdef USE_SPEAKER
+#include "esphome/components/speaker/speaker.h"
+#endif
+
+#include "esp_err.h"
+
+#include <freertos/FreeRTOS.h>
+
+namespace esphome {
+namespace audio {
+
+class AudioTransferBuffer {
+  /*
+   * @brief Class that facilitates tranferring data between a buffer and an audio source or sink.
+   * The transfer buffer is a typical C array that temporarily holds data for processing in other audio components.
+   * Both sink and source transfer buffers can use a ring buffer as the sink/source.
+   *   - The ring buffer is stored in a shared_ptr, so destroying the transfer buffer object will release ownership.
+   */
+ public:
+  /// @brief Destructor that deallocates the transfer buffer
+  ~AudioTransferBuffer();
+
+  /// @brief Returns a pointer to the start of the transfer buffer where available() bytes of exisiting data can be read
+  uint8_t *get_buffer_start() const { return this->data_start_; }
+
+  /// @brief Returns a pointer to the end of the transfer buffer where free() bytes of new data can be written
+  uint8_t *get_buffer_end() const { return this->data_start_ + this->buffer_length_; }
+
+  /// @brief Updates the internal state of the transfer buffer. This should be called after reading data
+  /// @param bytes The number of bytes consumed/read
+  void decrease_buffer_length(size_t bytes);
+
+  /// @brief Updates the internal state of the transfer buffer. This should be called after writing data
+  /// @param bytes The number of bytes written
+  void increase_buffer_length(size_t bytes);
+
+  /// @brief Returns the transfer buffer's currently available bytes to read
+  size_t available() const { return this->buffer_length_; }
+
+  /// @brief Returns the transfer buffers allocated bytes
+  size_t capacity() const { return this->buffer_size_; }
+
+  /// @brief Returns the transfer buffer's currrently free bytes available to write
+  size_t free() const;
+
+  /// @brief Clears data in the transfer buffer and, if possible, the source/sink.
+  virtual void clear_buffered_data();
+
+  /// @brief Tests if there is any data in the tranfer buffer or the source/sink.
+  /// @return True if there is data, false otherwise.
+  virtual bool has_buffered_data() const;
+
+  bool reallocate(size_t new_buffer_size);
+
+ protected:
+  /// @brief Allocates the transfer buffer in external memory, if available.
+  /// @return True is successful, false otherwise.
+  bool allocate_buffer_(size_t buffer_size);
+
+  /// @brief Deallocates the buffer and resets the class variables.
+  void deallocate_buffer_();
+
+  // A possible source or sink for the transfer buffer
+  std::shared_ptr<RingBuffer> ring_buffer_;
+
+  uint8_t *buffer_{nullptr};
+  uint8_t *data_start_{nullptr};
+
+  size_t buffer_size_{0};
+  size_t buffer_length_{0};
+};
+
+class AudioSinkTransferBuffer : public AudioTransferBuffer {
+  /*
+   * @brief A class that implements a transfer buffer for audio sinks.
+   * Supports writing processed data in the transfer buffer to a ring buffer or a speaker component.
+   */
+ public:
+  /// @brief Creates a new sink transfer buffer.
+  /// @param buffer_size Size of the transfer buffer in bytes.
+  /// @return unique_ptr if successfully allocated, nullptr otherwise
+  static std::unique_ptr<AudioSinkTransferBuffer> create(size_t buffer_size);
+
+  /// @brief Writes any available data in the transfer buffer to the sink.
+  /// @param ticks_to_wait FreeRTOS ticks to block while waiting for the sink to have enough space
+  /// @return Number of bytes written
+  size_t transfer_data_to_sink(TickType_t ticks_to_wait);
+
+  /// @brief Adds a ring buffer as the transfer buffer's sink.
+  /// @param ring_buffer weak_ptr to the allocated ring buffer
+  void set_sink(const std::weak_ptr<RingBuffer> &ring_buffer) { this->ring_buffer_ = ring_buffer.lock(); }
+
+#ifdef USE_SPEAKER
+  /// @brief Adds a speaker as the transfer buffer's sink.
+  /// @param speaker Pointer to the speaker component
+  void set_sink(speaker::Speaker *speaker) { this->speaker_ = speaker; }
+#endif
+
+  void clear_buffered_data() override;
+
+  bool has_buffered_data() const override;
+
+ protected:
+#ifdef USE_SPEAKER
+  speaker::Speaker *speaker_{nullptr};
+#endif
+};
+
+class AudioSourceTransferBuffer : public AudioTransferBuffer {
+  /*
+   * @brief A class that implements a transfer buffer for audio sources.
+   * Supports reading audio data from a ring buffer into the transfer buffer for processing.
+   */
+ public:
+  /// @brief Creates a new source transfer buffer.
+  /// @param buffer_size Size of the transfer buffer in bytes.
+  /// @return unique_ptr if successfully allocated, nullptr otherwise
+  static std::unique_ptr<AudioSourceTransferBuffer> create(size_t buffer_size);
+
+  /// @brief Reads any available data from the sink into the transfer buffer.
+  /// @param ticks_to_wait FreeRTOS ticks to block while waiting for the source to have enough data
+  /// @return Number of bytes read
+  size_t transfer_data_from_source(TickType_t ticks_to_wait);
+
+  /// @brief Adds a ring buffer as the transfer buffer's source.
+  /// @param ring_buffer weak_ptr to the allocated ring buffer
+  void set_source(const std::weak_ptr<RingBuffer> &ring_buffer) { this->ring_buffer_ = ring_buffer.lock(); };
+};
+
+}  // namespace audio
+}  // namespace esphome
+
+#endif