With your Access Codes created on iCap, your setup continues in EEG Cloud. This is where you create and run LEXI instances — the live jobs that generate your Live captions, Translation captions, and Voice. The instances deliver their output to your encoder over iCap, using the Access Codes you set up in Part 1.
You create one instance per caption service:
- A LEXI Live instance produces your primary captions.
- A LEXI Translate instance produces the translated captions and the LEXI Voice audio (S2–S6).
This guide uses S1 for English (the primary captions) and S2 for Spanish (the translation and voice).
Before you begin
Everything on this page is mapped to Access Codes created in Part 1 - UHD592 iCap Setup. If a code does not exist, is not bound to your encoder, or is not shared with EEGASR, the instance will be created but will not produce output. Confirm all four conditions below before you start.
| Check | Where to look | Correct state |
| EEG Cloud sign in | https://eegcloud.tv/login | You can sign in with your AI-Media Cloud Services username and password |
| Subscriptions active | Home Dashboard | LEXI Live and LEXI Translate both appear under Active Subscriptions |
| S1 access code | iCap Admin > Access Codes | The S1 code exists and shows Primary Encoder in green text |
| S2 access code | iCap Admin > Access Codes | The S2 code exists, is shared with EEGASR, and is on the same encoder as S1 |
Note: If LEXI Translate does not appear under Active Subscriptions, your account is not licensed for Translation or Voice. Contact AI-Media support before continuing.
Step 1: Create an API Key
The LEXI/iCap server on your encoder authenticates with an API key generated from your EEG Cloud account.
Note: Only users with permission to create API keys can complete this step. Your group's admin sets these permissions for each user. Users with the Group Admin role have this permission by default.
- Select your username in the top navigation bar, then select API Keys.
- On the Manage API Keys page, select Add New.
- Complete the Create a New API Key form:
- Product: select UHD592.
-
Serial Number: enter the serial number of your encoder.
- Select Submit. The API Key Created Successfully window shows your new API key.
-
Select Copy to Clipboard, then store the key somewhere secure. You need this key when you set up the encoder.
Important: This is the only time the API key is shown. If you close the window without saving it, you must create a new key.
- Select I've Saved the Key to close the window.
Step 2: Start LEXI Live for Primary Captions
LEXI Live generates the primary captions that everything downstream builds on.
- On the dashboard, select LEXI Live.
-
On the LEXI Live page, select Add New.
-
Complete the Create A New LEXI Instance form
- Name — a unique name to identify the instance.
- Access Code — the S1 Access Code set up from iCap.
- Engine — the ASR engine (This example uses LEXI 3.0).
- Language — the source caption language (This example is English (US)).
- Speaker Change Style — If a speaker change applies to your content, you can select how speaker changes are marked.
- Topic Model — optional, if one applies to your content.
Create A New LEXI Instance Form
- Check Advanced Options and confirm two settings before submitting. The rest of the advanced fields can stay at their defaults.
- Output Mode: must match the CC Output Format set on your encoder. The default is 608/708, which pairs with CTA 608/708 on the UHD592.
-
CC Service: set this to the service this instance feeds. The form states the rule directly: use 1 for CC1/S1 and 2 for CC3/S2. For the primary captions, S1.
Advanced Options - Create A New LEXI Instance Form
- Select Submit to save and toggle the State as On.
Step 3: Start LEXI Translate and Voice Instances
LEXI Translate produces the translated captions and, with a LEXI Voice subscription, the voice audio. It reads the primary captions produced in Step 2, so the LEXI Live instance must be running before the translation has anything to work from.
- Return to the dashboard and select LEXI Translate.
- On the LEXI Translate page, select Add New.
- Complete the Create A New LEXI Translate Instance form, then select Submit.
- Name — a unique name to identify the instance.
-
Access Code — the S2 Access Code set up from iCap (the service that carries the translated captions and voice).
- Translate From — the source language (This example is English).
-
Translate To — the destination language output (This example is Spanish).
-
LEXI Voice — the voice engine, selected from the dropdown. This appears only with a LEXI Voice subscription, and selecting it reveals the voice options below.
-
Default Voice Pool — select at least one voice. Output varies by speaker — explore and try different voices to match your desired result.
- Source CC Service — Select the CC Service (This example is CC1)
-
Destination Service — Select the Service within your CC Service (This example is S2 - must match the iCap Access Code Service)
- Select Submit to save and toggle the State as On.
On the UHD592, LEXI Voice starts only once the LEXI Translate instance for that Access Code is running here in EEG Cloud. Starting the process is an EEG Cloud action. The encoder has no control that starts Translate, so enabling voice on the encoder while this instance is Off produces no audio and no error. Start the instance here first, then configure the encoder.
Important: Source CC Service and Destination Service must line up with both the iCap Access Code and the encoder. The destination service you select here is the service the encoder expects the translation on.
Step 4: Validate the instances are running
Both instances are created as reusable profiles. Creating one does nothing on its own: the State toggle is what starts the job.
- For each instance, confirm the State toggle is On for the LEXI Live instance and the LEXI Translate instance.
- Open each instance and check Caption Activity. A running instance reports:
- Translate Status as Connected
- iCap Status of Translating
- WPM (Words per Minute) has a value greater than zero
What comes next
Leave both instances running before you move to the encoder. The UHD592 reads the EEG Cloud state rather than driving it, so the encoder assumes the instance State is already On.
With both instances running, continue to the encoder guides: