Voice Lab

PersonaPlex Audio Console

True NVIDIA nvidia/personaplex-7b-v1 audio-to-audio test surface on voice.beergram.net.

Open Native Studio API Docs OpenAPI
Checking service health...

Backend

-

Upstream

-

Mode

Native Audio WebSocket

Last Refresh

-

One-Click Latency Benchmark

Runs a live PersonaPlex websocket probe and reports connect, handshake, first-audio, and audio-loop roundtrip latency.

No benchmark run yet.

Connect

-

Handshake

-

First Audio

-

Loop Roundtrip

-

Live Audio Studio

Pop Out

Allow microphone access, select a voice prompt, and speak. This is the official PersonaPlex web client proxied through this console.

Realtime Resource Usage

Auto-refresh every 3s
Waiting for GPU metrics...
Container Status CPU Memory Network Block I/O
Waiting for container stats...

Quick API Map

Audio WS

GET wss://voice.beergram.net/api/chat?text_prompt=...&voice_prompt=NATF0.pt

Binary websocket passthrough to PersonaPlex upstream.

n8n Voice Relay

POST https://voice.beergram.net/v1/voice/relay

HTTP endpoint that uses true binary websocket framing internally.

Resource Metrics

GET https://voice.beergram.net/api/resources

GPU snapshot + container CPU, RAM, network, and block I/O.

Adapter Chat

POST https://voice.beergram.net/v1/chat

Legacy text contract for n8n; true PersonaPlex mode requires audio websocket.

Detailed endpoint docs: /api-docs and OpenAPI: /docs.