diff --git a/README.md b/README.md
index 836d2b6..3843d3f 100644
--- a/README.md
+++ b/README.md
@@ -12,8 +12,9 @@ optionally) in the background.
   other apps (you'll see an ongoing notification).
 - **Pause / resume** — while recording, a pause button appears next to the
   stop button.
-- **Stop** — finalizes the recording as an AAC `.m4a` file (small files,
-  playable everywhere) named like `recording_2026-07-09_14-30-00.m4a`.
+- **Stop** — finalizes the recording as an AAC `.m4a` file named like
+  `recording_2026-07-09_14-30-00.m4a`. Encoding is tuned for speech — mono,
+  22.05 kHz, 32 kbps — so an hour of journalling is roughly 14 MB.
 - **Automatic cloud upload** — when you stop, the file is queued with
   WorkManager and uploaded to your **Google Drive** (or an AWS S3 bucket if
   you switch backends). If you're offline, the upload waits for connectivity
diff --git a/app/src/main/java/com/audiojournal/app/recording/AudioRecorder.kt b/app/src/main/java/com/audiojournal/app/recording/AudioRecorder.kt
index 27eda64..a3ccf5d 100644
--- a/app/src/main/java/com/audiojournal/app/recording/AudioRecorder.kt
+++ b/app/src/main/java/com/audiojournal/app/recording/AudioRecorder.kt
@@ -41,8 +41,11 @@ class MediaRecorderAudioRecorder(private val context: Context) : AudioRecorder {
         recorder.setAudioSource(MediaRecorder.AudioSource.MIC)
         recorder.setOutputFormat(MediaRecorder.OutputFormat.MPEG_4)
         recorder.setAudioEncoder(MediaRecorder.AudioEncoder.AAC)
-        recorder.setAudioSamplingRate(44_100)
-        recorder.setAudioEncodingBitRate(128_000)
+        // Voice-oriented settings: a spoken-word journal does not need music
+        // fidelity, and these keep a one-hour entry around 14 MB instead of 58 MB.
+        recorder.setAudioChannels(1)
+        recorder.setAudioSamplingRate(22_050)
+        recorder.setAudioEncodingBitRate(32_000)
         recorder.setOutputFile(output.absolutePath)
         try {
             recorder.prepare()