CodeInformationActiveCity
LocalTalker
A Blueprint-first Unreal plugin for local character conversations with on-device llama.cpp generation and Qwen3-TTS speech.
Importance
84/100
Founding Year
2025
Region
Dialogue Marches
Terrain
Forest
Prestige
8
Activity
8
Stability
6
Reach
4
Stars
0
Forks
0
Contributors
1
Releases
0
Overview
LocalTalker is a late-era systems city focused on real-time local NPC conversation inside Unreal Engine. It combines in-process llama.cpp text generation, a persistent Qwen3-TTS worker, and game-facing conversation orchestration into one plugin-shaped product.
What makes it notable is that it is not just a model wrapper. The project includes the player interaction layer, turn-taking and proximity grouping, local voice pipeline management, and automation tests needed to make spoken character interaction feel like an actual gameplay system.
Buildings
- Builds a Blueprint-first Unreal plugin instead of a standalone AI demo.
- Runs local text generation and local speech synthesis as a game-facing conversation stack.
- Includes proximity grouping, turn-taking, player interaction routing, and automation coverage.
Production Queue
- Keep tightening latency, interruption handling, and spoken turn flow inside Unreal.
- Expand the packaging and setup path so the plugin is easier to move between projects.
- Push further on believable local character interaction as a reusable gameplay system.
Tech Tree
- Unreal Engine 5.7
- C++
- Blueprints
- llama.cpp
- Qwen3-TTS
- Python workers
Diplomacy
- Game developers
- AI builders
- Toolsmiths
- Unreal teams
Trade Routes
- Local NPC dialogue
- Speech synthesis
- Gameplay interaction
- Plugin architecture
Wonders
- Blueprint-first local conversation stack
- On-device NPC speech pipeline
Great Works
Landmark
On-Device Dialogue Director
A local conversation stack for Unreal characters that streams tokens, speaks replies, and manages who talks when.
Civilopedia Entry
Facet Type
Code
Repo
Owner: BigOtis
Name: LocalTalker
Role
Founder and primary engineer
Stack
Unreal Engine 5.7, C++, Blueprints, Python, llama.cpp, Qwen3-TTS
Features
Local character conversations in Unreal, Streaming token generation with in-process llama.cpp, Persistent TTS worker with Qwen3-TTS, Conversation director and player interaction components
Integrations
GitHub, llama.cpp, Qwen3-TTS
Architecture Notes
Treats local character conversation as a reusable Unreal gameplay subsystem rather than a chat prototype., Pairs runtime components with automation tests, voice-path configuration, and packaged-worker tooling.