diff --git a/csdr/chain/clientaudio.py b/csdr/chain/clientaudio.py
index 5441bc8..b67ace5 100644
--- a/csdr/chain/clientaudio.py
+++ b/csdr/chain/clientaudio.py
@@ -1,16 +1,18 @@
 from csdr.chain import Chain
-from pycsdr.modules import AudioResampler, Convert, AdpcmEncoder
+from pycsdr.modules import AudioResampler, Convert, AdpcmEncoder, Limit
 from pycsdr.types import Format
 
 
 class ClientAudioChain(Chain):
     def __init__(self, format: Format, inputRate: int, clientRate: int, compression: str):
         workers = []
-        if format != Format.FLOAT:
-            workers += [Convert(format, Format.FLOAT)]
         if inputRate != clientRate:
-            workers += [AudioResampler(inputRate, clientRate)]
-        workers += [Convert(Format.FLOAT, Format.SHORT)]
+            # we only have an audio resampler for float ATM so if we need to resample, we need to convert
+            if format != Format.FLOAT:
+                workers += [Convert(format, Format.FLOAT)]
+            workers += [AudioResampler(inputRate, clientRate), Limit(), Convert(Format.FLOAT, Format.SHORT)]
+        elif format != Format.SHORT:
+            workers += [Convert(format, Format.SHORT)]
         if compression == "adpcm":
             workers += [AdpcmEncoder(sync=True)]
         super().__init__(*workers)