Audio Generation

Script-First Approach

Before any audio is generated, you always review your script first. This lets you catch issues, tweak dialogue, and make sure everything reads the way you want — before spending credits.

Once your script looks good, you choose what to generate.

Generation Scopes

You don't have to generate everything at once. Pick the scope that fits your workflow:

Before generating, you can configure each character's voice — choose a preset, design a new voice, or use a reference audio clone.

Cost Estimation

Before you commit, you'll see an estimated credit cost based on your script's length and the scope you selected. No surprises.

What Happens During Generation

Here's what goes on behind the scenes:

1. Voices are generated first — each character's dialogue is produced with their assigned voice.

2. Timing is reconciled — the system adjusts the script's timing estimates to match the actual voice durations.

3. SFX and music are generated in parallel — sound effects and background music are created simultaneously, timed to fit the voices.

Real-Time Progress

You don't have to wait in the dark. As each segment completes, you'll see the status update live on screen. You can start listening to finished segments while the rest are still generating.

When Narra works on an edit for you — such as adding a segment, updating effects, or regenerating a part — the chat shows live progress so you can see what is running, what succeeded, and what needs attention.

Regeneration

Not happy with a specific line or sound? You have options:

• Regenerate a single segment — re-roll one voice line, sound effect, or music cue.

• Regenerate by speaker — regenerate all segments for a specific character at once.

• Add or revise segments through chat — ask Narra to add a new voice line, background music cue, or sound effect without manually rebuilding the whole script.

Each regeneration uses credits, but it's much cheaper than regenerating the entire project.

Limits

Each project supports up to 10 audio generation jobs. A job is one full generation run (e.g., generating all voices, or regenerating SFX). Individual segment regenerations do not count toward this limit.