diff --git a/packages/doorman-client/README.md b/packages/doorman-client/README.md
index 1e609af..ec11aeb 100644
--- a/packages/doorman-client/README.md
+++ b/packages/doorman-client/README.md
@@ -27,4 +27,16 @@ This will simulate a buzzer calling from 6133163433 (the test buzzer)
 
 If you let it poll for 10s then it should respond with Twilio xml saying to dial fallback numbers.
 
-If doorman-api is not running, it would always return with a `<Reject/>`
\ No newline at end of file
+If doorman-api is not running, it would always return with a `<Reject/>`
+
+## adding new audio assets
+
+Generate using this website: https://ttsmp3.com/
+
+select US English / Salli for consistency with the voice.
+
+After generated, go to this website: https://www.mp3louder.com/
+
+boost the audio 3db and save it
+
+put it in the doorman-client/assets folder with suffix `.protected.mp3`
\ No newline at end of file