Settings

These settings control the conversational AI experience in a simulation: microphone input, speech recognition (NLU), text-to-speech and the virtual assistant. In edit mode, you can change the following:

Setting	Description
Enable Microphone Input	Enable or disable microphone input. Must be enabled for conversational experiences.
Microphone Always Listening	Leave enabled for conversational experiences.
NLU Engine	Turns what you say in natural language into something a machine can understand. Authors can choose between Key Phrase (default) and **GPT.
TTS Engine	Turns text into spoken words. Authors can choose between Google and **ElevenLabs.
Virtual Person Language	The spoken language of all speaking entities. You can choose between English, Spanish, and *Hindi.
Virtual Assistant Avatar	A dropdown of virtual persons that have been added to the experience in the Scene Layout tab. Controls which VP should animate when using the AI Chat feature.
Virtual Assistant Voice	A dropdown of voices. These options are the same as listed in Natural Conversations.

*For Hindi, non-English text may not render correctly on the HyperSkill platform.

**Only available for paying/select HyperSkill authors.

NLU Engine

The NLU Engine option lets you choose between two options:

1. Key Phrase

   • Looks for and identifies important words in speech. The words or phrases must be an exact match. Great at detecting special keywords or phrases in speech.

2. GPT

   • High-accuracy technology capable of extrapolating from samples and understanding the flow and meaning of your words. The words or phrases do not need to be exact. This option is available for paying users.

TTS Engine

The TTS Engine option lets you choose between two options:

• Google

  • Generates text to speech using Google's Text-To-Speech technologies. Choose this option for quick and free speech.

• ElevenLabs

  • Generates high-quality speech using ElevenLabs' proprietary Text-To-Speech technologies. Choose this option to generate high-quality voices. This option is available for paying users.

Both engines also support custom voice IDs. See Natural Conversations.