What's the realtime factor for the Transcribe Streaming SDK?

Hi there!

We are using AWS Transcribe for streaming speech-to-text transcription. We're keen to understand whether the transcription is faster than realtime, and if so, what's the expected maximum factor?

As an example, if we buffer 5 seconds of audio data on the client before pushing into the stream for transcription, then send the buffer all at once, how long would we expect to wait before it catches up to real-time again? It seems, from experiments, that it's not faster than realtime (exactly 1x?), and transcription results stay at a 5 second delay as we push more audio data into it, but I wanted to verify that's the case!

Thank you very much, Ben