Software Downloads

Capacity Private Cloud is deployed using Kubernetes and Helm charts, providing a scalable, containerized architecture for speech recognition (ASR), text-to-speech (TTS), and voice biometrics capabilities. This article provides guidance on accessing platform software, developer SDKs, and integration tools.

Platform Deployment

The platform is distributed as Kubernetes Helm charts, enabling streamlined deployment and management of the complete speech technology stack.

Helm Charts Repository: https://github.com/lumenvox/helm-charts

To add the Helm repository and deploy:

helm repo add lumenvox https://lumenvox.github.io/helm-charts
helm repo update
helm search repo lumenvox
helm install lumenvox lumenvox/lumenvox -f my-lumenvox-values.yaml -n lumenvox --create-namespace

Prerequisites:

  • Kubernetes 1.33 or earlier
  • Helm 3+
  • External dependencies: PostgreSQL, MongoDB, Redis, and RabbitMQ

Minimum Testing Environment:

  • 3 nodes with 8+ CPUs and 16GB RAM each
  • MongoDB: 16GB memory, 300GB storage
  • PostgreSQL: 8GB memory, 30GB storage
  • RabbitMQ: 1 CPU, 2GB memory
  • Redis: 5GB memory

For detailed installation guidance, see the Getting Started section of this Knowledge Base.

Developer Integration

For developers building voice-enabled applications, the platform provides SDKs and APIs to integrate speech capabilities into your solutions.

Go SDK

The Go SDK provides a comprehensive client library for building voice-enabled applications. It handles gRPC connectivity, session management, audio streaming, and result processing.

Repository: https://github.com/lumenvox/go-sdk

Installation:

go get github.com/lumenvox/go-sdk

Requirements: Go 1.26 or later, plus access to a Capacity Private Cloud API endpoint.

Supported Capabilities:

  • Speech Recognition: ASR and transcription (streaming, batch, continuous)
  • Speech Synthesis: Text-to-speech with inline text or SSML URLs
  • Language Understanding: NLU operations including summarization and sentiment analysis
  • Audio Analysis: Answering machine detection, call progress analysis, speaker identification, and language detection
  • Text Processing: Normalization, redaction, grammar parsing, and intent detection

The repository includes 32 complete examples covering transcription, ASR, TTS, NLU tasks, and specialized analyses like diarization and call screening.

MRCP Connectivity

For organizations requiring MRCP (Media Resource Control Protocol) integration with existing telephony infrastructure, the MRCP API provides a Docker-based media server that bridges MRCP operations with the Capacity Private Cloud API.

Repository: https://github.com/lumenvox/mrcp-api

Quick Start:

cd docker
docker-compose up -d

Key Configuration:

  • MEDIA_SERVER__DEPLOYMENT_ID - Your deployment ID
  • MEDIA_SERVER__NUM_CHANNELS - Concurrent channel capacity (default: 200)
  • MEDIA_SERVER__SERVER_IP - Host machine IP for client routing
  • MEDIA_SERVER__LUMENVOX_API_ADDRESS - API service endpoint

Features:

  • OAuth support for secure API connections
  • Management API for runtime control (logging, maintenance mode)
  • Audio recording with save-waveform functionality
  • Custom certificate support for encrypted communications

The MRCP API enables seamless integration with media servers such as FreeSWITCH, Asterisk, and commercial platforms that support MRCP.

Additional Resources

For technical support inquiries, see When to Contact Support.

Was this article helpful?