Falcon 2 On-Premise Deployment

Benefits

With Falcon 2 On-Premise, you can:

• Keep all text and audio processing within your private infrastructure.
• Maintain full control over your voice data, infrastructure, and security policies.
• Meet security, compliance, and data residency requirements.
• Achieve ultra-low latency speech generation with sub-100 ms performance.
• Scale deployments across multiple servers or cloud instances as your workload grows.
• Retain complete control over encryption, network access, and operational management.

Deployment Options

Private Data Center

Deploy Falcon 2 on your organization's physical servers for maximum control and predictable performance.

Private Cloud (VPC)

Deploy Falcon 2 within your own Virtual Private Cloud (VPC) on AWS, Azure, GCP, OCI, or another supported cloud provider. Your data remains within your environment while benefiting from cloud scalability and availability.

A Falcon 2 on-premise deployment includes the following components:

1.  Murf Local License Agent (LLA):

An on-premises licensing proxy that enables secure licensing and authorization within your network.

2. Murf TTS Server

The core speech synthesis engine responsible for generating audio from text with low latency and scalable performance.

 

 Set Up LLA

The Murf Local License Agent (LLA) is an on-premises licensing proxy that enables secure text-to-speech generation within your network.

By deploying the LLA locally, you can:

• Reduce latency by processing requests within your infrastructure
• Keep voice data within your network perimeter
• Meet data privacy and compliance requirements
• Continue operating even when external connectivity is limited
  
  

Before You begin:

Hardware Requirements

• 1–2 vCPU minimum
• 1.5 GiB RAM minimum
• 5 GB available storage
  
  

Software Requirements

• Linux x86_64 (Ubuntu 20.04+, Debian 11+, RHEL 8+, or equivalent)
• Docker 20.10 or later
• Docker Compose 2.0 or later

Network Requirements

Ensure the server has outbound HTTPS (port 443) access to Murf licensing services. No public inbound ports are required.

Verify Docker Installation

Run the following commands to confirm Docker and Docker Compose are installed:

docker --version 
docker compose version 
docker run --rm hello-world 

Install and Deploy LLA

Step 1: Extract the Distribution Package

tar -xzf lla-distribution-latest.tar.gz 
cd lla-distribution-latest 

Step 2: Load Docker Images

./scripts/load-images.sh 

Verify the images are available:

docker images | grep -E 'local-license-agent|nginx' 

Step 3: Start the LLA Service

cd docker 
docker compose up -d 

Step 4: Verify the Deployment

Check that all containers are running:

docker compose ps 

Wait until the services show a healthy status.

Step 5: Test the Health Endpoint

curl http://localhost:8000/health 

A successful response indicates that the LLA is running and licensed correctly.

Step 6: Record the LLA Endpoint

You'll need the LLA endpoint when setting up the TTS Server.

• If LLA and TTS server are deployed on the same machine, use:

localhost:8000 

• If deployed on separate machines, use the hostname or IP address of the LLA server.
  
  

Set Up Falcon 2 TTS Server

The Falcon 2 TTS Server is the core engine of Murf's on-premises text-to-speech solution. It converts text into speech locally within your infrastructure, providing low-latency performance, enhanced privacy, and full operational control.

By deploying the Falcon 2 TTS Server locally, you can:

• Generate speech without relying on external APIs
• Keep text and audio data within your infrastructure
• Achieve low-latency speech generation
• Scale workloads using your own compute resources
• Maintain full control over security and access management

Ensure your server meets the following requirements:

Hardware Requirements

• NVIDIA GPU with at least 8 GB VRAM
• 4 vCPUs
• 16 GB RAM
• 50 GB SSD storage

Software Requirements

• Ubuntu 22.04 LTS or Amazon Linux 2023
• Docker Engine 20.10 or later
• NVIDIA GPU Drivers
• NVIDIA Container Toolkit

Pre-Deployment Checks

Before proceeding, verify that:

• NVIDIA GPU drivers are installed and functioning.
• Docker is installed and running.
• NVIDIA Container Toolkit is installed.
• Docker can access the GPU.
• You have received the Falcon 2 TTS Server Docker image from the Murf team.

Deploy the TTS Server

Step 1: Pull the Docker Image

docker pull <docker_image_url> 

Step 2: Configure Environment Variables

export TTS_MASTER_SECRET="your-secure-secret-key" 
export LLA_ENDPOINT="http://localhost:8000" 

 

If the TTS Server and LLA are hosted on different machines, replace the LLA endpoint with the appropriate hostname or IP address.

Step 3: Start the TTS Server

docker run -d \   
--gpus all \   
-e TTS_MASTER_SECRET="${TTS_MASTER_SECRET}" \   
-e LLA_ENDPOINT="${LLA_ENDPOINT}" \   
--name murf-tts \   
-p 80:8000 \   
--restart unless-stopped \   
<docker_image_url> 

 

Step 4: Verify the Deployment

docker ps 

Verify the Service

Check Container Logs

View the server logs to confirm successful startup:

docker logs -f murf-tts 

Test the Service

Open the following URL in your browser:

http://<server-ip>/ 

A successful response confirms that the TTS server is running.

Access API Documentation

To explore available endpoints, open:

http://<server-ip>/docs