Review the following setup instructions from Microsoft. They are quite clear and short.
General points:
1. Setup the Speech-to-text service (Microsoft Cognitive Services Speech Services) in an Azure instance. Your instance of the Speech-to-text service will be hosted in this instance. If privacy regulations are a concern, be sure to select an Azure instance that is in the same region as your Dynamics 365 instance.
2. Once setup, click 'Quick Start' and copy Key 1 and note the region indicated in the endpoint URL. Use these pieces of information to setup the reference in Dynamics 365. Note that when setting up the reference in Dynamics 365 all it needs is Key 1 and the region name spelled the same was as in the URL (eg: use need 'canadacentral' from the endpoint URL - https://canadacentral.api.cognitive.microsoft.com/sts/v1.0/issuetoken)
3. It is easy to confuse the two entities in Dynamics 365. Be sure to setup the 'Global' entity configuration first (point 2 above).
4. Note that when setting up a field on a entity for Speech-to-text that only first field setup for the entity will be used. I would recommend only setting up one field per entity.
5. Speech-to-text doesn't work on the iPhone. There are some noted workarounds to activate the microphone on an iPhone but it only allows one to record voice as a wave file. Instead use the native Dictation service. See my other blog on this.
