Skip to content
Agientry
← Back to Videos

PersonaPlex Enables Real-Time Full-Duplex Voice Conversation

Loading video
Loading video

NVIDIA PersonaPlex-7B uses a dual-stream speech model to listen and speak concurrently, generating low-latency replies while text role prompts and audio conditioning preserve a consistent persona.

Voice AIFull-Duplex SpeechSpeech-to-SpeechPersonaPlexNVIDIAMoshiDual-Stream ArchitecturePersona ControlVoice Interaction
Category: voice-interaction
Author: @liambraus
Date: 2026-10-22T00:00:00
Duration: 30.826s
Source: https://x.com/liambraus/status/2113234985918361742
Reference: https://github.com/NVIDIA/personaplex

Related Videos

ElevenLabs CLI v1: Voice Workflows in the Terminal
3:10

ElevenLabs CLI v1: Voice Workflows in the Terminal

@ElevenLabsDevs · 15 views · 2026-08-25
ElevenLabsCLI
Nemotron 3 Diarization tested: 100M-param real-time speaker diarization, better than expected locally
0:54

Nemotron 3 Diarization tested: 100M-param real-time speaker diarization, better than expected locally

@ouchi · 614 views · 2026-09-27
NVIDIANemotron 3
LiveKit Inference now runs cascaded, turn based and full duplex voice agents
1:06

LiveKit Inference now runs cascaded, turn based and full duplex voice agents

@livekit · 110 views · 2026-10-15
LiveKitVoice Agent
ElevenLabs introduces the Video to Music API
2:04

ElevenLabs introduces the Video to Music API

@ElevenLabsDevs · 99 views · 2026-10-15
ElevenLabsVideo to Music
BaoCut: Open-Source AI Bilingual Subtitle Translation
5:56

BaoCut: Open-Source AI Bilingual Subtitle Translation

@dotey · 124 views · 2026-10-14
BaoCutSubtitle Translation
Fluid Speech Mini: Open-Source Dictation ASR Under 300 MB, On-Device
0:46

Fluid Speech Mini: Open-Source Dictation ASR Under 300 MB, On-Device

@FluidVoiceApp · 159 views · 2026-10-13
Speech RecognitionASR

As an Amazon Associate, we earn from qualifying purchases.