Voice
LSDE
integrates a feature to manage your project's soundtrack, synchronized with your dialogues.
Prerequisites
You will first need to obtain an
Elevenlabs
API key.
Elevenlabs currently offers a free monthly plan, ideal for testing their API and discovering their expertise.
Configuration
To configure your project's voice system, go to the
voice configuration section
.
Define the voice export folder, where voices will be saved each time the project is saved.
The system allows you to name the output files.
This option adapts names to the needs of your game engine or project.
A drag-and-drop feature adjusts the order of labels.
Associate Variables (Automatic)
This example shows how to configure the voice system for a JRPG with multiple characters per dialogue, where your game engine needs to extract them.
Select a variable group for your speakers.
Check the "Multiple speakers per dialogue" option.
A section will open to configure a regular expression (`regex`) with a mandatory single group:
Group
id: Captures the value of the
tagto be searched for in the selected group's variables.
Group
value: Text captured for the
idgroup.
Optional, as you can write the text.
This group maximizes automation and reduces workload.
You will then see all the variables from the configured group, with their respective icon if provided.
Test your
regex
You can test your
regexto better understand it on
regexr
.
regex:
xxxxxxxxxjavascriptxxxxxxxxxxxxxxxx xxxxxxxxxxxxxx xxxxxxxxxxxxx
For example, by selecting the second capture.
Access the 'Details'.
You will notice that:
Group 1 captures your character's `tag id`
a1
.
Group 2 captures the rest of the text for the `id`
a1
.
Variables
In the
variables section
:
In the previously created
ACTORSgroup.
Providing the appearance greatly facilitates the organization and validation of dialogues and character personalities.
Note that you can modify this information at any time or complete it later.
Configure the Character
Let's return to
voice configuration
.
Associate a voice ID with your variable.
Voice IDs are created on the Elevenlabs platform.
Your Elevenlabs account already includes default models.
You will also be able to test and adjust the voice generator.
Create a Voice ID
On your
elevenlabs
profile:
Create a new custom voice.
LSDE
supports all API models.
I strongly recommend, at present, the
V3model for its emotional tag management, offering better narrative control.
After creation, go to your voices section.
Generate Voice Text
Once the ID is entered, to visualize or generate voice texts, open the
voice manager
window.
Select a key in the tree structure.
Do not choose a folder, as voice texts are displayed there in read-only mode.
The
regexwill capture the dialogue characters and offer a voice generation interface.
During generation, the text receives a "signature" at time T.
If the text changes without modifying the character order, an alert will prompt you to verify the need to regenerate the voice.
Associate Variables (Manually)
For scenarios with a single character per dialogue (without interactions), typical of simple projects.
You can uncheck the "Multiple speakers per key" box.
Configure Metadata
Associate each dialogue with the character(s) for whom you want to generate voices.
Select the key of the targeted dialogue.
Check each actor to manage for this dialogue.
Why multiple characters?
This can be useful for generic dialogues, such as a welcome phrase reused by several random NPCs with different personalities or genders.
Reorganization
Always on a key, not a folder.
The voice manager may signal a reorganization issue.
In this example, we have switched from an automatic assignment of multiple characters to a manual system.
If the system detects major changes in the order or number of characters, you will need to manually reorganize each already generated voice instance.
You just need to drag and drop each dialogue voice file to the associated character, declared in the metadata or via the automatic assignment system.