Audio - Cloosphere Guide

Admin › Settings › Audio

From the top, the screen is laid out as STT Settings → TTS Settings → Avatar.

STT (Speech-to-Text)

Configure the engine that converts user speech input into text.

Setting	Description	Default
Speech-to-Text Engine	STT service selection	`Whisper (Local)`

Transcribes with the built-in faster-whisper model on the server. This is the default and needs no external API key.

Setting	Description
STT Model	Whisper model name to use. If it is not on the server it is downloaded first

Setting	Description
API Key	Deepgram API key
STT Model	Optional — leave empty to use the default model

Setting	Description
API Key	Azure AI Speech Service key
Azure Region	Azure region (e.g., `koreacentral`)
Language Locales	Recognition languages, comma separated (e.g., `en-US,ja-JP`). Leave blank for auto-detect

Setting	Description	Default
GCP Project ID	Google Cloud project ID	-
Location	Region	`global`
STT Model	Cloud Speech-to-Text V2 model name (e.g., `chirp_2`)	-
Language Codes	Recognition languages (e.g., `ko-KR,en-US`)	`auto`
Service Account Key (Optional)	Leave empty to use Application Default Credentials	-
Use Global Google Cloud Key	Use the shared key registered on the server instead	-

Configure the engine that converts AI responses into speech.

Setting	Description	Default
Text-to-Speech Engine	TTS service selection	`Web API`
Response splitting	How a message is split for TTS requests — `Punctuation` (by sentence) · `Paragraphs` (by paragraph) · `None` (whole message as one string)	`Punctuation`

Uses the browser’s built-in Web Speech API. Only a TTS Voice needs to be chosen; no key is required.

Setting	Description
TTS Model	CMU ARCTIC speaker embedding name

Setting	Description
API Base URL	OpenAI API Endpoint
API Key	API authentication key
TTS Voice	`alloy`, `echo`, `fable`, `onyx`, `nova`, `shimmer`, etc.
TTS Model	e.g., `tts-1`, `tts-1-hd`

Setting	Description
API Key	Azure AI Speech Service key
Azure Region	Azure region (e.g., `koreacentral`)
TTS Voice	Voice selection (e.g., `ko-KR-SunHiNeural`)
Output format	Audio output format. The Available list link next to the label shows every format

Setting	Description
Language Filter (Optional)	Language used to narrow the voice list (e.g., `ko-KR, en-US`)
Service Account Key (Optional)	Leave empty to use Application Default Credentials
Use Global Google Cloud Key	Use the shared key registered on the server instead
TTS Voice	Voice selection

Setting	Description	Default
Location	Region	`us-central1`
Service Account Key (Optional)	Leave empty to use Application Default Credentials	-
Use Global Google Cloud Key	Use the shared key registered on the server instead	-
TTS Voice	Gemini TTS voice selection	-
TTS Model	Gemini TTS model	`gemini-2.5-flash-preview-tts`

Configure the avatar shown alongside AI responses.

Setting	Description	Default
Avatar Engine	Avatar service selection — `Default (None)` when unused	`Default (None)`

Selecting Azure AI Speech reveals the following fields.

Setting	Description
API Key	Avatar service authentication key
Azure Region	Service region (e.g., `eastus`, `koreacentral`)
Avatar Character	Avatar character selection
Avatar Style	Once a character is chosen, the styles that character supports are listed
Greeting Message	The first greeting shown when the avatar appears. Leave blank to use the default