Say, Pi dropping audio

[rosscado]

The SayPi state machine will "drop" some audio segments, meaning they don't get transcribed.

This occurs when:

• the state machine is in a "transcribing" state
• a "userStoppedSpeaking" event is received

The machine transitions to a "transcribing" state (self-transition) but does not exit or enter itself, so the transcribeAudio entry action is skipped.