nemotron-voice-agent-deploy

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Nemotron Voice Agent Deployment

Nemotron Voice Agent 部署

Real-time conversational AI voice agent using NVIDIA NIMs (ASR, TTS, LLM) with WebRTC (default) or WebSocket transport.

基于NVIDIA NIMs（ASR、TTS、LLM），通过WebRTC（默认）或WebSocket传输实现的实时对话式AI语音Agent。

Deployment Flow

部署流程

Always verify hardware first, even if user mentions a specific platform.

无论用户提及何种特定平台，务必先验证硬件情况。

STEP 1: Hardware Detection

步骤1：硬件检测

bash

nvidia-smi --query-gpu=name,memory.total --format=csv,noheader 2>/dev/null

Result	Action
Command fails / No output	→ Cloud NIMs
GPU detected	→ STEP 2: Platform Detection

bash

nvidia-smi --query-gpu=name,memory.total --format=csv,noheader 2>/dev/null

检测结果	操作建议
命令执行失败/无输出	→ Cloud NIMs
检测到GPU	→ 步骤2：平台检测

Cloud NIMs (No GPU)

Cloud NIMs（无GPU环境）

bash

cd nemotron-voice-agent
git submodule update --init
cp config/env.example .env

Export your NVIDIA API key:

bash

export NVIDIA_API_KEY=your-api-key  # Get from https://build.nvidia.com

Then edit

.env

bash

NVIDIA_LLM_MODEL=nvidia/nemotron-3-nano-30b-a3b  # Cloud model name

If user requests WebSocket transport, also add to

.env

bash

TRANSPORT=WEBSOCKET

bash

docker compose up --build --no-deps -d python-app ui-app

bash

cd nemotron-voice-agent
git submodule update --init
cp config/env.example .env

导出你的NVIDIA API密钥：

bash

export NVIDIA_API_KEY=your-api-key  # Get from https://build.nvidia.com

然后编辑

.env

文件：

bash

NVIDIA_LLM_MODEL=nvidia/nemotron-3-nano-30b-a3b  # 云端模型名称

若用户需要WebSocket传输，请在

.env

中添加以下内容：

bash

TRANSPORT=WEBSOCKET

bash

docker compose up --build --no-deps -d python-app ui-app

WebRTC: http://localhost:9000

WebSocket: http://localhost:7860/static/index.html


> **Note:** Deployment may take 30-60 minutes on first run.

**If user requests Multilingual mode**, also add to `.env`:
```bash
ENABLE_MULTILINGUAL=true
ASR_CLOUD_FUNCTION_ID=71203149-d3b7-4460-8231-1be2543a1fca
ASR_MODEL_NAME=parakeet-rnnt-1.1b-unified-ml-cs-universal-multi-asr-streaming

Remote Access:

ssh -L 9000:localhost:9000 user@host

http://<HOST_IP>:9000


> **注意：**首次部署可能需要30-60分钟。

**若用户需要多语言模式**，请在`.env`中添加以下内容：
```bash
ENABLE_MULTILINGUAL=true
ASR_CLOUD_FUNCTION_ID=71203149-d3b7-4460-8231-1be2543a1fca
ASR_MODEL_NAME=parakeet-rnnt-1.1b-unified-ml-cs-universal-multi-asr-streaming

**远程访问：**使用

ssh -L 9000:localhost:9000 user@host

或直接访问

http://<HOST_IP>:9000

STEP 2: Platform Detection (if GPU detected)

步骤2：平台检测（若检测到GPU）

bash

uname -m  # x86_64 → Workstation, aarch64 → Jetson
cat /etc/nv_tegra_release 2>/dev/null && echo "Jetson"

Platform	Reference	Requirements
Workstation (x86_64)	workstation-deployment.md	2x GPU (24GB+ VRAM), NIM containers
Jetson Thor (aarch64)	jetson-deployment.md	JetPack 7.0, Nemotron Speech ASR and TTS, vLLM

Note: Multilingual mode available on Workstation with WebRTC transport only.

bash

uname -m  # x86_64 → 工作站, aarch64 → Jetson
cat /etc/nv_tegra_release 2>/dev/null && echo "Jetson"

平台	参考文档	要求
Workstation (x86_64)	workstation-deployment.md	2块GPU（显存24GB以上）、NIM容器
Jetson Thor (aarch64)	jetson-deployment.md	JetPack 7.0、Nemotron Speech ASR和TTS、vLLM

**注意：**多语言模式仅在使用WebRTC传输的工作站环境中可用。