diff --git a/hugvey/client.py b/hugvey/client.py
index 0f9fe28..3b5a6ef 100644
--- a/hugvey/client.py
+++ b/hugvey/client.py
@@ -49,12 +49,13 @@ class VoiceServer(object):
         self.hugvey = hugvey
         
         self.chunk = 4096
-        self.mic_prerol_sec = .05
-        self.prerol_frame_count = math.ceil(self.input_rate / self.chunk)
+        self.mic_prerol_sec = .2
+        self.prerol_frame_count = math.ceil((self.input_rate / self.chunk) * self.mic_prerol_sec)
         self.prerol_frames = collections.deque(maxlen = self.prerol_frame_count)
         
         self.p = pyaudio.PyAudio()
         # wait a sec for the input devices to come up
+        logger.debug("Use a mic prerol of {} frames".format(self.prerol_frame_count))
         logger.debug('wait for mic')
         time.sleep(3)
         logger.debug('done waiting for mic')