diff --git a/campatmo.py b/campatmo.py
index fbcfebd..00ccac4 100755
--- a/campatmo.py
+++ b/campatmo.py
@@ -4,16 +4,83 @@ import sys
 import random
 from glob import glob
 from pygame import mixer
+from typing import Union, List, Mapping
 
 from atmorepl import AtmoReplRunner
 
 VERBOSITY = 0
 
 
+class SoundCollection:
+    def __init__(
+        self, name, files: Union[List[str], Mapping[str, List[str]]], volume=0.5
+    ):
+        self.name = name
+        self.topics = files.keys() if hasattr(files, "keys") else []
+        if self.topics:
+            self.files = files
+        else:
+            self.files = {"None": files}
+        self.flat_files = [f for topic in self.files for f in self.files[topic]]
+        self.volume = volume
+        self.cur_topic = None
+        self.load_all_files()
+        self.sound_indices = {name: 0 for name in self.sounds}
+        self.flat_sound_index = 0
+
+    def load_all_files(self):
+        self.sounds = {}
+        if self.topics:
+            for t in self.topics:
+                self.load_topic_files(t)
+        else:
+            self.load_topic_files("None")
+        self.flat_sounds = [s for topic in self.sounds for s in self.sounds[topic]]
+        random.shuffle(self.flat_sounds)
+
+    def load_topic_files(self, topic):
+        self.sounds[topic] = []
+        for f in self.files[topic]:
+            sound = mixer.Sound(f)
+            sound.set_volume(self.volume)
+            self.sounds[topic].append(sound)
+        random.shuffle(self.sounds[topic])
+
+    def next(self):
+        if self.cur_topic is None:
+            if len(self.flat_sounds) == 0:
+                return
+            self.flat_sound_index += 1
+            if self.flat_sound_index >= len(self.flat_sounds):
+                random.shuffle(self.flat_sounds)
+                self.flat_sound_index = 0
+            return self.flat_sounds[self.flat_sound_index]
+        else:
+            if len(self.sounds[self.cur_topic]) == 0:
+                return
+            self.sound_indices[self.cur_topic] += 1
+            if self.sound_indices[self.cur_topic] >= len(self.sounds[self.cur_topic]):
+                random.shuffle(self.sounds[self.cur_topic])
+                self.sound_indices[self.cur_topic] = 0
+            return self.sounds[self.cur_topic][self.sound_indices[self.cur_topic]]
+
+    def set_topic(self, topic=None):
+        if self.topics:
+            self.cur_topic = topic
+
+    def __len__(self):
+        if self.cur_topic is None:
+            return len(self.flat_sounds)
+        else:
+            return len(self.sounds[self.cur_topic])
+
+
 class CampAtmo:
     tracks_per_type = 2
     single_track_types = ["lines"]
-    sound_volume = 0.5
+    background_track_types = ["background", "buzzwords"]
+    lines_delta_sec = (5, 15)
+    line_reaction_wait_sec = 0.8
 
     def __init__(self):
         mixer.init(frequency=22050 * 2)
@@ -21,7 +88,17 @@ class CampAtmo:
 
     def load_sounds(self):
         type_names = glob("samples/*/")
-        type_files = {c.split("/")[1]: glob(c + "/*.ogg") for c in type_names}
+        type_files = {}
+        for type_dir in type_names:
+            type_name = type_dir.split("/")[1]
+            flat_files = glob(type_dir + "/*.ogg")
+            type_files[type_name] = flat_files
+            topic_files = glob(type_dir + "/*/*.ogg")
+            if topic_files:
+                type_files[type_name] = {}
+                for topic_dir in glob(type_dir + "/*/"):
+                    topic_name = topic_dir.split("/")[2]
+                    type_files[type_name][topic_name] = glob(type_dir + "/*/*.ogg")
         self.types = {}
         i = 0
         mixer.set_num_channels(
@@ -42,39 +119,26 @@ class CampAtmo:
                 i += 1
         self.sounds = {}
         for name, files in type_files.items():
-            self.sounds[name] = []
-            for f in files:
-                sound = mixer.Sound(f)
-                sound.set_volume(self.sound_volume)
-                self.sounds[name].append(sound)
-            random.shuffle(self.sounds[name])
-
-    def manage_type_queue(self, name, type_channels):
-        try:
-            next_sound_index = max(
-                [
-                    self.sounds[name].index(tt.get_queue() or tt.get_sound())
-                    for tt in type_channels
-                ]
-            )
-        except ValueError:
-            next_sound_index = 0
-        if next_sound_index >= len(self.sounds[name]):
-            random.shuffle(self.sounds[name])
-            next_sound_index = 0
+            self.sounds[name] = SoundCollection(name, files)
+
+    def manage_background_queue(self, name, type_channels):
         for c in type_channels:
             if c.get_queue() is not None and len(self.sounds[name]) > 1:
                 continue
-            c.queue(self.sounds[name][next_sound_index])
+            v(name)
+            c.queue(self.sounds[name].next())
             if c.get_queue() is None and len(self.sounds[name]) > 1:
-                c.queue(self.sounds[name][next_sound_index + 1])
+                c.queue(self.sounds[name].next())
 
     def run_forever(self):
         while True:
             for name, type_channels in self.types.items():
-                if len(self.sounds[name]) == 0 or name in self.single_track_types:
+                if (
+                    name not in self.background_track_types
+                    or len(self.sounds[name]) == 0
+                ):
                     continue
-                self.manage_type_queue(name, type_channels)
+                self.manage_background_queue(name, type_channels)
             time.sleep(1)
             self.load_sounds()
 
diff --git a/samples/laughter/.exists b/samples/reactions/.exists
similarity index 100%
rename from samples/laughter/.exists
rename to samples/reactions/.exists