diff --git a/mozilla_voice_tts/bin/synthesize.py b/mozilla_voice_tts/bin/synthesize.py
index 3de97555..1e2886b8 100644
--- a/mozilla_voice_tts/bin/synthesize.py
+++ b/mozilla_voice_tts/bin/synthesize.py
@@ -155,7 +155,11 @@ if __name__ == "__main__":
     if args.gst_style is None:
         gst_style = C.gst['gst_style_input']
     else:
-        gst_style = args.gst_style
+        # check if gst_style string is a dict, if is dict convert  else use string
+        try:
+            gst_style = json.loads(args.gst_style)
+        except:
+            gst_style = args.gst_style
 
     wav = tts(model, vocoder_model, args.text, C, args.use_cuda, ap, use_griffin_lim, args.speaker_fileid, speaker_embedding=speaker_embedding, gst_style=gst_style)