[New feature] Speech - Speech-To-Text processing

[nrobert]

Ability to process Speech-To-Text from:

• a file input
• a microphone

It could be by using a normal or custom speech model

The demo should contain the following setup:

• Speech resource region
• Speech key
• Language selection
• (optional) custom speech endpoint id (and if provided, no language selection as custom speech endpoints are only valid for 1 language)