🎛️ Part of the TizWildin Plugin Ecosystem — 19 free audio plugins with a live update dashboard.

FreeEQ8 · XyloCore · Instrudio · Therum · BassMaid · SpaceMaid · GlueMaid · MixMaid · MultiMaid · MeterMaid · ChainMaid · PaintMask · WURP · AETHER · WhisperGate · RiftWave · FreeSampler · VF-PlexLab · PAP-Forge-Audio

🎁 Free Packs & Samples — jump to free packs & samples

🎵 Awesome Audio — (FREE) Awesome Audio Dev List

VocalForge PersonaPlex Lab

VocalForge PersonaPlex Lab is a GitHub-ready starter repo for building a JUCE plugin + local backend + HTML tester around NVIDIA PersonaPlex, aimed at voice-conditioned vocal asset generation, spoken hook design, whispers, chants, robotic/system phrases, and sample-pack export.

It is designed around one hard rule:

Never run PersonaPlex inference on the plugin audio thread.

Instead, this repo separates concerns cleanly:

• JUCE plugin for DAW-side control, auditioning, and export workflow
• Python backend bridge for queued jobs, health checks, and filesystem outputs
• HTML tester for fast iteration outside the DAW
• PersonaPlex dependency layer tracked separately through upstream or your fork

Why this repo exists

NVIDIA PersonaPlex is a real-time, full-duplex speech-to-speech conversational model with text-based role prompts and audio-based voice conditioning. It supports a fixed set of NAT and VAR voice prompts, a local server mode, and an offline WAV-in/WAV-out path. The upstream code is MIT licensed, while the model weights are released under the NVIDIA Open Model License. citeturn909810view0

That makes PersonaPlex a strong foundation for a vocal phrase forge, but not something you should bolt directly into a DAW's audio callback. This repository gives you the product shell around it.

Product framing

Best framing for this project:

• AI vocal phrase generator
• vocal sample pack builder
• whisper / chant / spoken phrase lab
• offline voice-conditioned export tool
• JUCE + browser control surface for local voice generation

Not the best framing yet:

• live note-played vocal synth
• transparent replacement for human singers
• real-time sample-accurate instrument engine

Upstream PersonaPlex facts this repo builds around

Upstream PersonaPlex provides:

• a local HTTPS server launched with python -m moshi.server --ssl ...
• an offline evaluator launched with python -m moshi.offline
• fixed voice prompt IDs including NATF/NATM/VARF/VARM sets
• support for HF_TOKEN after the user accepts the model license on Hugging Face
• optional CPU offload via accelerate

The upstream README describes PersonaPlex as a real-time, full-duplex speech-to-speech model based on the Moshi architecture and lists the NAT/VAR voice sets and offline/server commands. citeturn909810view0

Recommended repo strategy

Use this repository as your main product repo.

Keep PersonaPlex itself separate:

• use upstream directly if you are only consuming it
• use your fork if you need engine-level changes
• mount it as a submodule, subtree, or local path dependency

Suggested layout:

VocalForge_PersonaPlex_Lab/
  plugin/                  # JUCE plugin shell
  backend/                 # FastAPI bridge + queue + runners
  tester/                  # HTML frontend tester
  shared/                  # schemas, voice maps, shared metadata
  docs/                    # architecture, packaging, attribution, SEO, roadmap
  scripts/                 # dev helpers, linkage scripts, repo setup
  third_party/
    personaplex/           # optional git submodule / local checkout / your fork

Features in this starter repo

Included now

• JUCE plugin scaffold with async backend client structure
• FastAPI backend with job queue and placeholder WAV output
• HTML tester that talks to the same backend
• JSON schema and voice mapping starter files
• GitHub templates, issue forms, CI, docs, release notes scaffolding
• clear attribution + dependency strategy for PersonaPlex

Intended next wiring steps

• replace placeholder generation with real PersonaPlex invocation
• connect plugin UI to waveform preview and local library browser
• add drag-export, batch render, and pack metadata export
• package the backend for Catalina / Windows shipping
• decide on submodule vs fork path for PersonaPlex integration

Quickstart

1) Clone this repo

git clone <your-repo-url>
cd VocalForge_PersonaPlex_Lab

2) Add PersonaPlex as a dependency

Option A: local clone/fork placed under third_party/personaplex

git clone https://github.com/NVIDIA/personaplex third_party/personaplex

Option B: submodule

git submodule add https://github.com/NVIDIA/personaplex third_party/personaplex

3) Set up the backend

cd backend
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
uvicorn app:app --reload --port 8765

4) Run the tester

cd tester
python3 -m http.server 8080

5) Point the backend runner to PersonaPlex

Edit:

• backend/runners/personaplex_runner.py

This is where you can choose one of three integration modes:

• spawn python -m moshi.offline
• call a running local PersonaPlex server
• build a thin Python-native wrapper if stable in your environment

SEO / discoverability targets

This repo is intentionally structured for findability around terms like:

• JUCE AI vocal plugin
• PersonaPlex plugin wrapper
• local voice generation plugin
• AI sample pack generator
• NVIDIA PersonaPlex JUCE integration
• offline vocal phrase generator
• speech-to-speech plugin scaffold
• DAW vocal asset generator

See docs/seo-keywords.md for the full keyword map and README strategy.

Topics to use on GitHub

Recommended topics:

• juce
• vst3
• audio-plugin
• speech-to-speech
• personaplex
• moshi
• fastapi
• html-tester
• voice-generation
• sample-pack-builder
• offline-ai
• local-ai
• audio-tools

Licensing and attribution

This repository contains your product scaffolding only.

• Your wrapper/plugin/backend code: choose the license you want for this repo
• NVIDIA PersonaPlex upstream code: MIT license upstream citeturn909810view0
• PersonaPlex model weights: NVIDIA Open Model License upstream citeturn909810view0

Read docs/upstream-attribution.md before redistributing any PersonaPlex-linked build or claiming generated-output rights.

Roadmap

• main repo scaffold
• backend queue and starter endpoints
• HTML tester shell
• JUCE project shell
• real PersonaPlex runner wiring
• packaged helper app
• waveform preview / library browser
• drag-to-DAW export
• batch phrase CSV pipeline
• release binaries and screenshots

Status

This is a serious starter repository, not a finished commercial product. It is intentionally shaped so you can layer PersonaPlex under it cleanly and keep upstream updates manageable.

TizWildin FREE sample packs

Pack	Description
TizWildin-Aurora	3-segment original synth melody pack with loops, stems, demo renders, and neon/cinematic phrasing
TizWildin-Obsidian	Dark cinematic sample pack with choir textures, menu loops, transitions, bass, atmosphere, drums, and electric-banjo extensions
TizWildin-Skyline	30 BPM-tagged synthwave and darkwave loops with generator snapshot and dark neon additions
TizWildin-Chroma	Multi-segment game synthwave loop sample pack from TizWildin Entertainment
TizWildin-Chime	Multi-part 88 BPM chime collection spanning glass, void, halo, reed, and neon synthwave lanes
Free Violin Synth Sample Kit	Physical-model violin sample kit rendered from the Instrudio violin instrument
Free Dark Piano Sound Kit	88 piano notes + dark/cinematic loops and MIDI
Free 808 Producer Kit	94 hand-crafted 808 bass samples tuned to every chromatic key
Free Riser Producer Kit	115+ risers and 63 downlifters - noise, synth, drum, FX, cinematic
Phonk Producer Toolkit	Drift phonk starter kit - 808s, cowbells, drums, MIDI, templates
Free Future Bass Producer Kit	Loops, fills, drums, bass, synths, pads, and FX

Related audio projects

• VF-PlexLab - VocalForge PersonaPlex Lab starter repo for a JUCE plugin + local backend + HTML tester around NVIDIA PersonaPlex.
• PAP-Forge-Audio - Procedural Autonomous Plugins runtime for generating, branching, validating, and restoring plugin projects from natural-language sound intent.