Audio Channel Architecture and Call Bridge Framework
Overview
The platform is evolving toward a more modular and scalable voice architecture through the introduction of a new component called Call Bridge.
Several existing platform services continue to be leveraged, including:
Speech Services
Runtime Engine
AI-Vendors
Channel Gateway
The new Call Bridge module acts as a central audio integration layer that simplifies the process of connecting new voice channels and telephony providers to the platform.
Purpose of Call Bridge
Call Bridge consolidates the complex audio-processing and telephony-related functionality into a reusable framework. Instead of each audio integration implementing the entire voice stack independently, channel developers only need to implement a defined set of interfaces.
Key Benefits
Simplified onboarding of new voice channels
Reduced integration complexity
Consistent audio processing across providers
Reusable telephony capabilities
Easier expansion to future voice platforms
How the Integration Works
High-Level Call Flow
A customer call enters a Genesys Architect flow.
Architect invokes the Call Audio Connector action.
Genesys establishes a secure WebSocket connection with VNext.
Audio is streamed bi-directionally between Genesys and VNext.
VNext processes the conversation using:
Speech-to-Text (STT)
Natural Language Understanding (NLU)
Text-to-Speech (TTS)
VNext captures customer intent and relevant business information.
When the conversation ends, VNext returns the configured output variables to Genesys.
Architect resumes execution and routes the call to the appropriate agent queue or workflow.
Scalability Improvements
One of the primary goals of the new architecture is to improve scalability.
Current Challenge
Previous implementations faced limitations such as port restrictions that constrained environment size.
Expected Improvement
With the introduction of FreeSWITCH and the Call Bridge architecture:
Traditional port limitations can be removed
Environment capacity becomes primarily dependent on available infrastructure resources
Larger-scale deployments become possible
Audio Connector Framework
To integrate a new audio channel, developers must implement a standardized Phone Interface.
The interface allows an audio connector to advertise its supported capabilities to the Call Bridge framework.
Configure a Genesys Audio Connector Connection
Refer to the set-up document
Advertised Capabilities
An audio connector can specify support for:
Audio format support
Sample rate configuration
Outbound calling
Call transfer
Call recording
DTMF detection
DTMF playback
Audio playout tracking
These capabilities enable the platform to understand what functionality is available for a given voice channel.
Recommended Audio Standard
The platform is moving toward standardization on:
L16 Audio Format
Characteristics:
Linear PCM audio
16-bit audio format
Commonly used as a universal audio processing format within the platform
Standardizing on L16 reduces channel-specific implementation complexity and enables consistent speech processing.
Call Lifecycle Management
Once a voice interaction is established, the connector is responsible for creating a Call Object.
The Call Object provides:
Events
Inbound call events
Outbound call events
Call state updates
Audio stream events
Lifecycle Handling
Developers must correctly implement the required event handling and lifecycle behaviors to ensure reliable call processing.
After integration is complete, the connector gains access to platform voice capabilities, including:
Speech Recognition
Speech Synthesis
Real-time audio processing
Runtime Engine integrations
AI-driven conversational experiences
Audio Transcoding
Different voice providers and telephony networks use different codecs. To support interoperability, Call Bridge includes built-in transcoding capabilities.
Included Codec Support
The framework contains codec resamplers that can handle:
Sample-rate conversion
Example: 24 kHz → 8 kHz
μ-law (Mu-Law)
A-law
Developers may also:
Use their own codec libraries
Integrate third-party audio-processing libraries when needed
PSTN Compatibility
For Public Switched Telephone Network (PSTN) connectivity, audio may need to be transcoded from the platform's internal format (L16 PCM) to telephony codecs such as:
G.711
Other provider-specific formats
The Call Bridge framework provides the necessary foundation for these conversions while allowing extensions when additional codecs are required.
WhatsApp Voice Codec Considerations
Discussion during the session highlighted that WhatsApp voice services may support codecs such as:
Opus
AMR-related codecs
Other low-bandwidth voice formats
If additional codec support becomes necessary, the framework can be extended through supplementary codec libraries.
DTMF Support Requirements
A critical requirement for audio-channel implementations is proper DTMF handling.
Supported Scenarios
Receiving DTMF tones
Sending DTMF tones
IVR navigation
External telephony integrations
Important Consideration
Any configured codec must support reliable in-band DTMF transmission when required.
This is especially important when calls traverse:
PSTN networks
IVR systems
Third-party telephony platforms
Failure to support DTMF correctly may impact menu navigation and call-routing functionality.
Architectural Benefits
The introduction of Call Bridge provides significant advantages:
Simplified voice-channel development
Reusable telephony infrastructure
Reduced implementation effort
Consistent audio-processing pipeline
Scalable deployment model
Easier support for future voice providers
Standardized codec and audio handling
Direct access to platform speech and AI capabilities
Conclusion
The Call Bridge architecture serves as a unified telephony framework that abstracts the complexities of audio processing, codec management, call lifecycle handling, and channel integration. By implementing a small set of standardized interfaces, new voice channels can quickly leverage the platform's existing speech services, runtime engine, and AI capabilities, accelerating development while improving scalability, maintainability, and future extensibility.
Last updated
Was this helpful?