Skip to main content
Introduction to Sanas SDK and how it works

What is Sanas SDK?

Sanas SDK is a lightweight connector that enables real-time audio streaming via AI models hosted on the Sanas Cloud. The Sanas ecosystem consists of two components:
  • Connector (local Python library) - Installed on endpoint devices/servers, handles SIP authentication and RTC audio streaming
  • Sanas Cloud (hosted service) - Hosts AI models and provides real-time audio processing, enhances quality, and returns clean audio

Available AI Models

Sanas SDK provides two specialized models optimized for different audio processing needs:
Optimized for: Speech recognition systems (ASR)Use this model when:
  • Building voice assistants or chatbots
  • Running ASR in noisy environments
  • Need to improve Word Error Rate (WER)
  • Optimizing turn-taking in conversational AI
Key Benefits:
  • 5-30% WER improvement in noisy conditions
  • Works with any ASR (Deepgram, Google, Whisper, Azure, AWS)
  • Optimized for machine processing
Model Name: AGENTIC_NCConfiguration:
    audio_params.modelName = "AGENTIC_NC"
    audio_params.sampleRate = 16000
Learn more →

How It Works

Sanas SDK operates as a cloud-hosted real-time streaming service that processes audio with low latency:

Technical Architecture

Connection Protocol:
  • SIP (Session Initiation Protocol): Establishes and manages the connection
  • RTP (Real-Time Transport Protocol): Transmits audio streams
Processing Mode:
  • Real-time streaming
  • Audio processed in chunks as it arrives

How Audio Flows

1

Take audio input

Take an audio stream from your voice pipeline.
2

Establish SIP connection

Establish a secure SIP connection to the Sanas Cloud via the SDK connector.
3

Stream audio via RTP

Transmit audio chunks in real time over RTP.
4

Process with AI

Process each 20ms audio chunk with the selected AI model.
5

Receive clean audio stream

Receive enhanced audio in real time via RTP.
6

Use enhanced audio

Send enhanced audio to the next stage based on your use case.

Deployment Options


System Requirements

Supported Platforms

ComponentRequirement
Operating SystemUbuntu 22.04 x86-64
Python VersionPython 3.10 or higher
NetworkOutbound connectivity to Sanas Cloud (SIP/RTC ports)
DependenciesDependencies are handled automatically during installation.

Network Requirements

Required ports and protocols:
  • SIP: For connection establishment
  • RTP: For audio streaming
Firewall configuration: Ensure your firewall allows outbound SIP and RTP traffic to Sanas Cloud endpoints.

Key Features

Real-Time Streaming

Low-latency processingLive audio processing with SIP/RTC for minimal delay

High Concurrency

Scalable solutionProcess multiple audio streams simultaneously

Session Management

SIP protocolReliable connection establishment and management

ASR-Agnostic

Works with any ASRDeepgram, Google, Whisper, Azure, AWS, etc.

Easy Integration

Simple APIInitialize, create a processor, stream audio

Secure

Enterprise securityEncrypted transmission, secure authentication

Next Steps

Ready to start? Follow the quickstart guide to install the SDK and process your first live audio stream in under 10 minutes.

Need Help?