Project Management

VoiceLink does a lot of things automatically TTS, DSP, delivery, change detection. But before any of that runs, there's a more basic question: who's allowed to push the button?

This deep dive is about the management side of VoiceLink how access is controlled, how projects are organized, and how you keep track of what happened when something inevitably goes sideways.

It's not the part that looks good in a screenshot. But it's the part that decides whether your production scales or collapses under its own weight the moment a second studio gets involved.

User Management

Interacting with VoiceLink starts by not having access by default. You register, and an admin or project manager assigns you to a project.

They can even limit your accessible items to specific localizations and content packages. That way you can enforce the Principle of Least Privilege. Everyone sees exactly what they need and nothing more.

Content Packages

The term Content Package was already dropped a few times, but what exactly is it?
We use Content Packages to split your project into multiple smaller parts. That makes it more manageable and faster to run deliveries and process, since it only operates on the data assigned to that content package instead of the entirety of the data. 
That ensures fast iteration times and performance for larger projects.

How you split your project is up to you. You can have one big main content package and for every DLC of your game a separate content package or you split your project by missions/scenarios or whatever other factor that you can think of. 
The organisational side of that is completely up to you. 

The data in a content package is not exclusive. One character might be part of multiple content packages depending on how you split the project. This pretty much acts like a prefiltering. User's don't need to see the entirety of the VO data and can work in isolated manageable units.

In the following video you can see how easy it is to assign individual VoiceLines or entire Characters to a content package.

Roles & Permissions

VoiceLink takes a different starting point: people's access should match the slice of work they're responsible for, not the tool's all-or-nothing permission model.

So instead of a fixed set of roles, you build your own and each one can be scoped to exactly the work it covers. A few that come up often:

• A casting manager role that can assign voice actors or tts models to characters but can't trigger TTS generation.
• A Japanese localization lead role limited to the JP language and the content packages that touch it. They see the dialogue that's theirs to direct and nothing from another region.
• A contractor role scoped to a single content package one client's work, one batch of lines, and a clean wall everywhere else.

The scoping is the point. A role isn't just "what can you do" it's "what can you see," and that boundary follows the lines of the work itself.

A voice director who's also handling the EN localization pass gets both roles and their access is the union. When they step off the localization piece, you remove that one role the director role and everything it grants stays untouched. No re-provisioning from scratch, no "create another user," no accidental over-granting just to avoid the paperwork.

It lets you go as granular as a project demands without turning permission management into its own job. You define roles once, hand them out in combinations, and adjust as the project moves. 

Within each section of VoiceLink, access comes in three levels None, View, or Edit and you set them independently. Someone can view characters but not edit them. Someone else can edit content packages but not edit voice lines itself. Someone can kick off TTS generation but can't touch the final export pipeline. You can go as crazy as you like and put as much effort into it.

Small details, daily difference

The thing that makes a permission system survive day-to-day use isn't the policy engine it's whether you can see who's who at a glance. So every role gets a color and an icon.

It sounds minor until you're scanning a team of forty, trying to remember who has the Voice Designer role and who's contractors-only. A color-coded badge next to each name answers it in a second. You stop hunting through settings panels and start trusting the interface to reflect the structure you built. Roles stop being something you configure once and forget, they become a visible part of the pipeline.

Logs & Traceability

Every change in VoiceLink is logged and fully traceable.

No more questioning if someone ruined your beautiful VO line. One quick search and you know you're not going crazy it actually was different the day before. The whole history is right there including engine logs, papertrails and command runs, which are all connected to a user.

What if something goes horribly wrong?

Probably the most important question for any production: what happens when everything goes wrong? Someone deletes important recordings, or whole sections of finalized voice lines and characters vanish.

Normally that's a nightmare and a massive undertaking to restore. But since VoiceLink's data lives entirely in a database with regular backups and your files are handled by your version control, we can recover a previous state rather quickly. No digging through scattered folders or praying someone kept a local copy.