---
title: "Discussion on \"PersonaPlex: Voice and Role Control for Full Duplex Conversational Speech Models (2026)\" | Hashnode"
description: "Discussion on \"PersonaPlex: Voice and Role Control for Full Duplex Conversational Speech Models (2026)\". The Agent Text stream starts with an instruction prompt, while the Agent Audio stream starts with a voice sample. After that, previously generated text and audio are fed back in as inputs.\nWhile the m"
url: "https://hashnode.com/posts/personaplex-voice-and-role-control-for-full-duplex-conversational-speech-models-2026/6b00818fa74f77f612871c11"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [PersonaPlex: Voice and Role Control for Full Duplex Conversational Speech Models (2026)](https://nickonedrop.hashnode.dev/personaplex-voice-and-role-control-for-full-duplex-conversational-speech-models-2026?utm_source=hashnode&utm_medium=feed)

The Agent Text stream starts with an instruction prompt, while the Agent Audio stream starts with a voice sample. After that, previously generated text and audio are fed back in as inputs. While the m

1d ago · 2 min read · nickonedrop.hashnode.dev  
Read the article

-   [#deeplearning](https://hashnode.com/tag/deeplearning)
-   [#machine-learning](https://hashnode.com/tag/machine-learning)
-   [#voice-ai](https://hashnode.com/tag/voice-ai)
-   [#speech-models](https://hashnode.com/tag/speech-models)

## Written by

[Nick One-Drop](https://hashnode.com/@nickonedrop)

@nickonedrop

  
[View profile](https://hashnode.com/@nickonedrop)

ML researcher and solo builder behind monore. Creating things that make people feel.

ML researcher and solo builder behind monore. Creating things that make people feel.

## Responses

Markdown supported · ⌘↵ to post

No responses yet.
