diff --git a/client/src/output/audio.rs b/client/src/output/audio.rs
index 2e5124d..a555f2d 100644
--- a/client/src/output/audio.rs
+++ b/client/src/output/audio.rs
@@ -30,8 +30,26 @@ impl AudioOut {
             .expect("no src element")
             .downcast::<gst_app::AppSrc>()
             .expect("src not an AppSrc");
-
+        
+        src.set_caps(Some(&gst::Caps::builder("audio/mpeg")        // mpeg‑4 AAC
+            .field("mpegversion", &4i32)
+            .field("stream-format", &"adts")
+            .build()));
         src.set_format(gst::Format::Time);
+
+        {
+            let bus = pipeline.bus().expect("bus");
+            std::thread::spawn(move || {
+                use gst::MessageView::*;
+                for msg in bus.iter_timed(gst::ClockTime::NONE) {
+                    if let Error(e) = msg.view() {
+                        error!("💥 client‑audio: {} ({})",
+                               e.error(), e.debug().unwrap_or_default());
+                    }
+                }
+            });
+        }
+
         pipeline.set_state(gst::State::Playing)?;
 
         Ok(Self { src })
diff --git a/server/src/audio.rs b/server/src/audio.rs
index fd197dc..a6b2ff0 100644
--- a/server/src/audio.rs
+++ b/server/src/audio.rs
@@ -7,6 +7,7 @@ use gstreamer as gst;
 use gstreamer_app as gst_app;
 use gst::prelude::*;
 use gst::ElementFactory;
+use gst::MessageView::*;
 use lesavka_common::lesavka::AudioPacket;
 use tokio_stream::wrappers::ReceiverStream;
 use tonic::Status;
@@ -36,10 +37,10 @@ impl Drop for AudioStream {
 }
 
 /*───────────────────────────────────────────────────────────────────────────*/
-/*  eye_ear() – capture from ALSA (“speaker”) and push AAC AUs via gRPC     */
+/*  ear() – capture from ALSA (“speaker”) and push AAC AUs via gRPC          */
 /*───────────────────────────────────────────────────────────────────────────*/
 
-pub async fn eye_ear(alsa_dev: &str, id: u32) -> anyhow::Result<AudioStream> {
+pub async fn ear(alsa_dev: &str, id: u32) -> anyhow::Result<AudioStream> {
     //   NB: one *logical* speaker → id==0.  A 2nd logical stream could be
     //   added later (for multi‑channel) without changing the client.
     gst::init().context("gst init")?;
@@ -65,15 +66,14 @@ pub async fn eye_ear(alsa_dev: &str, id: u32) -> anyhow::Result<AudioStream> {
 
     let (tx, rx) = tokio::sync::mpsc::channel(8192);
 
-    let bus = pipeline.bus().expect("no bus");
+    let bus = pipeline.bus().expect("bus");
     std::thread::spawn(move || {
         for msg in bus.iter_timed(gst::ClockTime::NONE) {
-            use gst::MessageView::*;
             match msg.view() {
                 Error(e)   => error!("💥 audio pipeline: {} ({})",
-                                    e.error(), e.debug().unwrap_or_default()),
+                                      e.error(), e.debug().unwrap_or_default()),
                 Warning(w) => warn!("⚠️  audio pipeline: {} ({})",
-                                    w.error(), w.debug().unwrap_or_default()),
+                                      w.error(), w.debug().unwrap_or_default()),
                 StateChanged(s) if s.current() == gst::State::Playing =>
                     debug!("🎶 audio pipeline PLAYING"),
                 _ => {}
@@ -93,7 +93,7 @@ pub async fn eye_ear(alsa_dev: &str, id: u32) -> anyhow::Result<AudioStream> {
                     std::sync::atomic::AtomicU64::new(0);
                 let n = CNT.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
                 if n < 10 || n % 300 == 0 {
-                    debug!("🔊 eye‑ear #{n}: {} bytes", map.len());
+                    debug!("🎧 ear #{n}: {} bytes", map.len());
                 }
 
                 let pts_us = buffer
@@ -111,7 +111,7 @@ pub async fn eye_ear(alsa_dev: &str, id: u32) -> anyhow::Result<AudioStream> {
                         std::sync::atomic::AtomicU64::new(0);
                     let d = DROPS.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
                     if d % 300 == 0 {
-                        warn!("🔊💔 dropped {d} audio AUs (client too slow)");
+                        warn!("🎧💔 dropped {d} audio AUs (client too slow)");
                     }
                 }
                 Ok(gst::FlowSuccess::Ok)
@@ -129,17 +129,17 @@ pub async fn eye_ear(alsa_dev: &str, id: u32) -> anyhow::Result<AudioStream> {
 }
 
 fn build_pipeline_desc(dev: &str) -> anyhow::Result<String> {
-    use gst::ElementFactory;                           // <- simpler probe
+    // choose the first encoder that exists on the system
     let enc = ["voaacenc", "avenc_aac", "fdkaacenc"]
         .into_iter()
-        .find(|&e| ElementFactory::find(e).is_some())
+        .find(|e| ElementFactory::find(e).is_some())      // cheap run‑time probe
         .ok_or_else(|| anyhow::anyhow!("no AAC encoder plugin available"))?;
 
     Ok(format!(
-        // ➊ provide-clock=false lets the USB gadget be master
-        // ➋ audioconvert+audioresample make sure caps match encoder
+        // 48 kHz stereo, floats, gadget is master → provide‑clock=false
         "alsasrc device=\"{dev}\" provide-clock=false do-timestamp=true ! \
-         audioconvert ! audioresample ! audio/x-raw,channels=2,rate=48000,format=F32LE ! \
+         audioconvert ! audioresample ! \
+         audio/x-raw,format=F32LE,channels=2,rate=48000 ! \
          {enc} bitrate=192000 ! aacparse ! queue ! \
          appsink name=asink emit-signals=true max-buffers=64 drop=true"
     ))
diff --git a/server/src/main.rs b/server/src/main.rs
index bfcc470..206b2c0 100644
--- a/server/src/main.rs
+++ b/server/src/main.rs
@@ -192,7 +192,7 @@ impl Relay for Handler {
         let dev = std::env::var("LESAVKA_ALSA_DEV")
             .unwrap_or_else(|_| "hw:UAC2Gadget,0".into());
 
-        let s = audio::eye_ear(&dev, 0)
+        let s = audio::ear(&dev, 0)
             .await
             .map_err(|e| Status::internal(format!("{e:#}")))?;