Q

Quantized Dia 1.6B Int8

Developed by RobAgrees
Dia is a 1.6 billion parameter open-source text-to-speech model that supports highly realistic dialogue and non-verbal expression generation
Downloads 69
Release Time : 4/28/2025

Model Overview

Dia is a text-to-speech model developed by Nari Labs that directly generates highly realistic dialogue from text, supports emotion and tone control through audio input, and can produce non-verbal expressions such as laughter and coughing.

Model Features

Dynamic int8 Quantization
Utilizes dynamic quantization technology for lighter deployment and faster inference, improving inference speed by approximately 20%
Multi-speaker Dialogue Generation
Generates multi-character dialogues using [S1] and [S2] tags
Non-verbal Expression Support
Supports generating non-verbal expressions such as laughter, coughing, and throat clearing
Voice Cloning Functionality
Supports voice cloning through example code

Model Capabilities

Text-to-Speech
Multi-speaker Dialogue Generation
Non-verbal Expression Generation
Voice Cloning

Use Cases

Dialogue Systems
Virtual Assistants
Generates natural conversational speech for virtual assistants
Produces highly realistic dialogue effects
Game NPCs
Generates dynamic voice dialogues for game characters
Supports multi-character interactions and emotional expressions
Content Creation
Audio Content Production
Generates dialogue content for podcasts, audiobooks, etc.
Can produce natural dialogues including non-verbal expressions
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase