diff --git a/include/SDL3/SDL_audio.h b/include/SDL3/SDL_audio.h
index ac1540bc0..9b8737919 100644
--- a/include/SDL3/SDL_audio.h
+++ b/include/SDL3/SDL_audio.h
@@ -860,12 +860,13 @@ extern DECLSPEC int SDLCALL SDL_GetAudioStreamAvailable(SDL_AudioStream *stream)
 
 
 /**
- * Get the number of sample frames currently queued.
+ * Get the number of bytes currently queued.
  *
- * Since audio streams can change their input format at any time, even if
- * there is still data queued in a different format, this reports the queued
- * _sample frames_, so if you queue two stereo samples in float32 format and
- * then queue five mono samples in Sint16 format, this will return 6.
+ * Note that audio streams can change their input format at any time, even if
+ * there is still data queued in a different format, so the returned byte
+ * count will not necessarily match the number of _sample frames_ available.
+ * Users of this API should be aware of format changes they make when feeding
+ * a stream and plan accordingly.
  *
  * Queued data is not converted until it is consumed by
  * SDL_GetAudioStreamData, so this value should be representative of the exact
@@ -878,7 +879,7 @@ extern DECLSPEC int SDLCALL SDL_GetAudioStreamAvailable(SDL_AudioStream *stream)
  * clamped.
  *
  * \param stream The audio stream to query
- * \returns the number of sample frames queued.
+ * \returns the number of bytes queued.
  *
  * \threadsafety It is safe to call this function from any thread.
  *
diff --git a/src/audio/SDL_audiocvt.c b/src/audio/SDL_audiocvt.c
index a5be406db..891a425c4 100644
--- a/src/audio/SDL_audiocvt.c
+++ b/src/audio/SDL_audiocvt.c
@@ -660,7 +660,7 @@ int SDL_PutAudioStreamData(SDL_AudioStream *stream, const void *buf, int len)
     }
 
     if (retval == 0) {
-        stream->total_frames_queued += len / SDL_AUDIO_FRAMESIZE(stream->src_spec);
+        stream->total_bytes_queued += len;
         if (stream->put_callback) {
             const int newavail = SDL_GetAudioStreamAvailable(stream) - prev_available;
             stream->put_callback(stream->put_callback_userdata, stream, newavail, newavail);
@@ -863,7 +863,7 @@ static int GetAudioStreamDataInternal(SDL_AudioStream *stream, void *buf, int ou
             SDL_assert(!"Not enough data in queue (read)");
         }
 
-        stream->total_frames_queued -= output_frames;
+        stream->total_bytes_queued -= input_bytes;
 
         // Even if we aren't currently resampling, we always need to update the history buffer
         UpdateAudioStreamHistoryBuffer(stream, input_buffer, input_bytes, NULL, 0);
@@ -953,7 +953,7 @@ static int GetAudioStreamDataInternal(SDL_AudioStream *stream, void *buf, int ou
     if (SDL_ReadFromAudioQueue(stream->queue, input_buffer, input_bytes) != 0) {
         SDL_assert(!"Not enough data in queue (resample read)");
     }
-    stream->total_frames_queued -= input_frames;
+    stream->total_bytes_queued -= input_bytes;
 
     // Update the history buffer and fill in the left padding
     UpdateAudioStreamHistoryBuffer(stream, input_buffer, input_bytes, left_padding, padding_bytes);
@@ -1124,7 +1124,7 @@ int SDL_GetAudioStreamQueued(SDL_AudioStream *stream)
     }
 
     SDL_LockMutex(stream->lock);
-    const Uint64 total = stream->total_frames_queued;
+    const Uint64 total = stream->total_bytes_queued;
     SDL_UnlockMutex(stream->lock);
 
     // if this overflows an int, just clamp it to a maximum.
@@ -1142,7 +1142,7 @@ int SDL_ClearAudioStream(SDL_AudioStream *stream)
     SDL_ClearAudioQueue(stream->queue);
     SDL_zero(stream->input_spec);
     stream->resample_offset = 0;
-    stream->total_frames_queued = 0;
+    stream->total_bytes_queued = 0;
 
     SDL_UnlockMutex(stream->lock);
     return 0;
diff --git a/src/audio/SDL_sysaudio.h b/src/audio/SDL_sysaudio.h
index 980dbc565..e0d931606 100644
--- a/src/audio/SDL_sysaudio.h
+++ b/src/audio/SDL_sysaudio.h
@@ -179,7 +179,7 @@ struct SDL_AudioStream
     float freq_ratio;
 
     struct SDL_AudioQueue* queue;
-    Uint64 total_frames_queued;
+    Uint64 total_bytes_queued;
 
     SDL_AudioSpec input_spec; // The spec of input data currently being processed
     Sint64 resample_offset;