SpeechToText-WebSockets-Javascript | SDK & Sample to do speech recognition | Speech library

I'm trying to get the Bing Speech API to work in C# via WebSockets. I've looked through the implementation in Javascript here and have been following the protocol instructions here, but I've come up against a complete brick wall. I can't use the existing C# service because I'm running in a Linux container, so I need to use an implementation on .net Core. Annoyingly, the existing service is closed-source!

I can connect to the web socket successfully, but I can't ever get the server to respond to my connection. I'm expecting to receive a turn.start text message from the server, but I get booted off the server as soon as I've sent a few bytes of an audio file. I know the audio file is in the right format because I've got it directly from the C# service sample here.

I feel like I’ve exhausted the options here. The only thing I can think of now is that I’m not sending the audio chunks correctly. Currently, I’m just sending the audio file in consecutive 4096 bytes. I know the first audio message contains the RIFF header which is only 36 bytes, and then I'm just sending this along with the next (4096-36) bytes.

Here is my code in full. You should just be able to run it as a .net core or .net framework console application, and will need an audio file and an API key.