Just updated our CFD to be able to utilise this. We create a call flow based on reading a database which tells the system what prompts to play and where to redirect to for any given day or time of day. Normally we would need to record the audio and upload it as part of that step however now there is also the option for quick changes and short term messages to just say whatever is typed into the field as well. Works great! I'm going to look at allowing "next available slot" message to be updated directly from a web interface using it. Thanks. One thing that would be good - an option to cache the TTS for each component. Some TTS will only have a few variables or not change very often so if it could hold on to the Wav File from Amazon and record the Text/SSML that was sent to get it (or even a hash of it) then you could check that cache first for the file. This would give an offline backup if Amazon Polly was not reachable for any reason, speed up the connection and incur less usage charges from Amazon.