diff --git a/routes/listening_post.py b/routes/listening_post.py
index 44cf682..a094fbd 100644
--- a/routes/listening_post.py
+++ b/routes/listening_post.py
@@ -346,7 +346,7 @@ def scanner_loop():
         logger.info("Scanner thread stopped")
 
 
-def _start_audio_stream(frequency: float, modulation: str):
+def _start_audio_stream(frequency: float, modulation: str):
     """Start audio streaming at given frequency."""
     global audio_process, audio_rtl_process, audio_running, audio_frequency, audio_modulation
 
@@ -442,43 +442,51 @@ def _start_audio_stream(frequency: float, modulation: str):
             'pipe:1'
         ]
 
-        try:
-            # Use shell pipe for reliable streaming
-            # Log stderr to temp files for error diagnosis
-            rtl_stderr_log = '/tmp/rtl_fm_stderr.log'
-            ffmpeg_stderr_log = '/tmp/ffmpeg_stderr.log'
-            shell_cmd = f"{' '.join(sdr_cmd)} 2>{rtl_stderr_log} | {' '.join(encoder_cmd)} 2>{ffmpeg_stderr_log}"
-            logger.info(f"Starting audio: {frequency} MHz, mod={modulation}, device={scanner_config['device']}")
-
-            audio_rtl_process = None  # Not used in shell mode
-            audio_process = subprocess.Popen(
-                shell_cmd,
-                shell=True,
-                stdout=subprocess.PIPE,
-                stderr=subprocess.PIPE,
-                bufsize=0,
-                start_new_session=True  # Create new process group for clean shutdown
-            )
-
-            # Brief delay to check if process started successfully
-            time.sleep(0.3)
-
-            if audio_process.poll() is not None:
-                # Read stderr from temp files
-                rtl_stderr = ''
-                ffmpeg_stderr = ''
-                try:
-                    with open(rtl_stderr_log, 'r') as f:
-                        rtl_stderr = f.read().strip()
-                except:
-                    pass
-                try:
-                    with open(ffmpeg_stderr_log, 'r') as f:
-                        ffmpeg_stderr = f.read().strip()
-                except:
-                    pass
-                logger.error(f"Audio pipeline exited immediately. rtl_fm stderr: {rtl_stderr}, ffmpeg stderr: {ffmpeg_stderr}")
-                return
+        try:
+            # Use shell pipe for reliable streaming
+            # Log stderr to temp files for error diagnosis
+            rtl_stderr_log = '/tmp/rtl_fm_stderr.log'
+            ffmpeg_stderr_log = '/tmp/ffmpeg_stderr.log'
+            shell_cmd = f"{' '.join(sdr_cmd)} 2>{rtl_stderr_log} | {' '.join(encoder_cmd)} 2>{ffmpeg_stderr_log}"
+            logger.info(f"Starting audio: {frequency} MHz, mod={modulation}, device={scanner_config['device']}")
+
+            audio_rtl_process = None  # Not used in shell mode
+            audio_process = subprocess.Popen(
+                shell_cmd,
+                shell=True,
+                stdout=subprocess.PIPE,
+                stderr=subprocess.PIPE,
+                bufsize=0,
+                start_new_session=True  # Create new process group for clean shutdown
+            )
+
+            # Brief delay to check if process started successfully
+            time.sleep(0.3)
+
+            if audio_process.poll() is not None:
+                # Read stderr from temp files
+                rtl_stderr = ''
+                ffmpeg_stderr = ''
+                try:
+                    with open(rtl_stderr_log, 'r') as f:
+                        rtl_stderr = f.read().strip()
+                except:
+                    pass
+                try:
+                    with open(ffmpeg_stderr_log, 'r') as f:
+                        ffmpeg_stderr = f.read().strip()
+                except:
+                    pass
+                logger.error(f"Audio pipeline exited immediately. rtl_fm stderr: {rtl_stderr}, ffmpeg stderr: {ffmpeg_stderr}")
+                return
+
+            # Validate that audio is producing data quickly
+            try:
+                ready, _, _ = select.select([audio_process.stdout], [], [], 2.0)
+                if not ready:
+                    logger.warning("Audio pipeline produced no data in startup window")
+            except Exception as e:
+                logger.warning(f"Audio startup check failed: {e}")
 
             audio_running = True
             audio_frequency = frequency
@@ -878,6 +886,7 @@ def audio_debug() -> Response:
     """Get audio debug status and recent stderr logs."""
     rtl_log_path = '/tmp/rtl_fm_stderr.log'
     ffmpeg_log_path = '/tmp/ffmpeg_stderr.log'
+    sample_path = '/tmp/audio_probe.bin'
 
     def _read_log(path: str) -> str:
         try:
@@ -897,7 +906,34 @@ def audio_debug() -> Response:
         'audio_process_alive': bool(audio_process and audio_process.poll() is None),
         'rtl_fm_stderr': _read_log(rtl_log_path),
         'ffmpeg_stderr': _read_log(ffmpeg_log_path),
+        'audio_probe_bytes': os.path.getsize(sample_path) if os.path.exists(sample_path) else 0,
     })
+
+
+@listening_post_bp.route('/audio/probe')
+def audio_probe() -> Response:
+    """Grab a small chunk of audio bytes from the pipeline for debugging."""
+    global audio_process
+
+    if not audio_process or not audio_process.stdout:
+        return jsonify({'status': 'error', 'message': 'audio process not running'}), 400
+
+    sample_path = '/tmp/audio_probe.bin'
+    size = 0
+    try:
+        ready, _, _ = select.select([audio_process.stdout], [], [], 2.0)
+        if not ready:
+            return jsonify({'status': 'error', 'message': 'no data available'}), 504
+        data = audio_process.stdout.read(4096)
+        if not data:
+            return jsonify({'status': 'error', 'message': 'no data read'}), 504
+        with open(sample_path, 'wb') as handle:
+            handle.write(data)
+        size = len(data)
+    except Exception as e:
+        return jsonify({'status': 'error', 'message': str(e)}), 500
+
+    return jsonify({'status': 'ok', 'bytes': size})
 
 
 @listening_post_bp.route('/audio/stream')