AWS for M&E Blog

Conversational XR sports viewing experiences powered by Amazon Nova Sonic

This blog is co-authored by José Somolinos, Portfolio Manager – XR, AI & Innovation, Accedo, and Christian Marko, Chief Innovation Officer, Stats Perform.

An image showing the simulated in-headset view of the Accedo XR view including the soccer game play broadcast in the middle. The four sides of the game play have different information screens. At the top, the current score is shown (flanked on either side by the Accedo Xtend logo and a shoe ad), in-game statistics are on the left-hand side. A 3D image of the playing field is under the game play showing additional statistics along with a bar stating who the player in current control of the ball. The live Amazon Nova and Stats Perform speech-to-speech interface is on the right-hand side of the game play.

Figure 1: Accedo Xtend headset experience.

The convergence of cloud computing and advanced viewing technologies is revolutionizing how we experience sports and media content. Through extended reality (XR) headsets and smart glasses, viewers are stepping into a new era of immersive entertainment. These cutting-edge devices can transform the viewing experience by seamlessly blending real-time statistics, multiple camera perspectives, and 3D overlays while maintaining environmental awareness.

To make these XR experiences truly hands-free, they need natural voice interaction. However, traditional voice applications are complex, requiring multiple separate components for speech recognition, language processing, and audio output. This complexity can result in unnatural-sounding speech that loses important qualities, such as tone and speaking style.

We will describe how AWS Partners, Accedo and Stats Perform, used Amazon Nova Sonic to build an innovative XR sports viewing platform, featured during the 2025 International Broadcasting Convention (IBC) at the AWS booth.

Accedo and Stats Perform

Accedo is a global provider of software and services to optimize, grow, and evolve video streaming services. The offerings of Accedo range from no-code video solutions and immersive applications to complete over-the-top (OTT) managed services. They are backed by a data-driven approach and the end-user experience in mind.

Stats Perform is a global leader in sports artificial intelligence (AI) and data technology. They power richer fan experiences across major sports through Opta data. They provide access to their 7.2 petabytes of proprietary Opta sports data for broadcasters, websites, sportsbooks, leagues, teams, and more. The real-time data is delivered through various feeds and APIs, offering everything from ultrafast live scores to detailed player statistics, metrics, predictions, graphics, and derived insights.

Amazon Nova Sonic

The Amazon Nova Sonic model (offered by Amazon Web Services (AWS) and available in Amazon Bedrock) provides real-time, conversational interactions—both speech and text output. Through its Bidirectional Streaming API, the model enables near real-time, natural voice conversations with preserved contextual elements. Amazon Nova Sonic processes and responds to real-time speech as it occurs, enabling natural, human-like conversational experiences.

The real-time speech-to-speech persistent bidirectional connection allows simultaneous event streaming. Unlike traditional request-response patterns, this approach enables the following:

  • Continuous audio streaming from the user to the model
  • Concurrent speech processing and generation
  • Near real-time model responses without waiting for complete utterances

The Bidirectional Streaming API consists of three main components:

  1. Session initialization: The client establishes a bidirectional stream and sends the configuration events
  2. Audio streaming: The model processes incoming user audio that is captured, encoded, and streamed as events in near real-time
  3. Response streaming: As audio arrives, the model simultaneously sends event responses:
Diagram that explains the Amazon Nova Sonic bidirectional streaming system workflow. A user provides an audio chat to the Client Application. The Bidirectional Streaming API then determines the correct response stream to provide Amazon Bedrock understanding for use with Amazon Nova Sonic.

Figure 2: High-level overview of the Bidirectional Stream API.

Amazon Nova Sonic delivers a transformative approach to conversational AI with its unified speech understanding and generation architecture. This speech-to-speech foundation model (FM) helps enterprises build voice experiences that remain natural and contextually aware.

Accedo Xtend solution

Video 1: Accedo Xtend headset experience—visual demonstration only, no audio.

The Accedo Xtend solution is designed to bring immersive sports to life across augmented reality (AR) and virtual reality (VR) headsets such as the Meta Quest, Apple Vision Pro, and XREAL. The on-device Accedo Xtend experience offers fans a large virtual screen for watching live sporting events. It provides enriched real-time stats, automated replays, and multi-angle viewing—all managed dynamically through the Accedo Control content management system (CMS) on AWS.

A voice-enabled AI agent, powered by Amazon Nova Sonic, and developed in collaboration with Stats Perform, acts as a virtual co-commentator. Fans can interact with match insights and personalize their viewing experience. The modular setup makes it quick to adapt UI layouts, content, and Stats Perform Opta data in real time.

Accedo Xtend is opening new doors for monetization through ads and sponsorships, while redefining how fans connect with the game—whether at home or on the move. Users can access the XR application through AR or VR headsets to enjoy high-quality video and interactive, data-rich modules in an immersive environment.

A workflow diagram depicting how the content management system influences the layout and configuration of the Accedo Xtend solution shown on the device.

Figure 3: The Accedo Xtend solution.

The architecture: How the solution leverages AWS

A high-level architecture diagram detailing the AWS architecture and flow for this solution. A step-by-step explanation of the diagram's flow is detailed within the blog.

Figure 4: High-level AWS architecture and flow.

The following describes how Accedo, Stats Perform and AWS services power the XR viewing experience:

  1. Sports video feeds are captured, encoded, and transmitted to AWS.
  2. Incoming video streams can be transported using AWS Elemental MediaConnect for video transport. AWS Elemental MediaLive handles transcoding into an adaptive bitrate (ABR) ladder. AWS Elemental MediaPackage handles packaging and origination, and Amazon CloudFront is the content delivery network (CDN) that delivers the video stream. Stats Perform analyzes game feeds to capture metrics, analytics, predictive data, and visuals, and serves this information through APIs.
  3. The Stats Perform OptaAI Studio chat API utilizes the Anthropic Claude family of FMs in Amazon Bedrock to power a conversational chat experience.
  4. To power the generative AI user interface voice interaction module, Stats Perform integrates with Amazon Bedrock and the Amazon Nova Sonic FM to enable real-time, conversational interactions through the Bidirectional Streaming API.
  5. When initiating a conversation, Stats Perform incorporates the match context of the relevant game. Users can have an interactive voice experience about the live gameplay. The low-latency Amazon Nova Micro FM powers the dynamic user interface behavior. It provides guides for showing contextualized statistics, graphics or video related to the ongoing conversation.
  6. The on-device application connects to the CMS, streams video content from the CDN, and interfaces with Stats Perform APIs. Users can access the Accedo XR solution through their AR or VR headset to view both the game’s video projection and interactive statistics in an immersive environment.
A looping video showing the text of a back-and-forth conversation between the end-user and the AI agent. Following is the conversation. Note that image blurs are visible in the conversation and cover redacted information for the purpose of this example only and would not normally appear during a true user experience. User: Who scored during the game AI Agent: [image blur] scored a hat-trick for Leipzip! User: What is the current score now AI Agent: The current score is [image blur] Munich 0-3 [image blur] Leipzig. User: Who do you think will win the game AI Agent: Based on the performance, Leipzig seems to be in a strong position to win the game. At the base of the interaction screen, under the red Speech button, Accedo Extend indicates: I'm listening - ask me anything!

Video 2: Speech-to-speech interaction module inside the XR experience.

Integrating Amazon Nova Sonic

In order to utilize Amazon Nova Sonic, Stats Perform needed to create new prompt types because speech-based interactions differ from text-based ones. For instance, speech-to-speech differs from chat-based interactions because speech may be limited in context. These interactions also require back-and-forth exchanges to gather more information before moving the conversation forward. Additionally, prompts that generate long-form speech outputs create a poor user experience, as users must spend time listening to find the correct answer.

For the solution, Stats Perform architected a workflow to leverage their knowledge base to deliver accurate statistics and results. The user can engage in a conversation directly with the AI agent. The new Stats Perform system prompts accept the external inputs to create a truly unique and scalable customer experience. The AI agent can flexibly serve different applications and use cases depending on the customer’s needs.

We have additional guidance on system prompts, using the Bidirectional Streaming API, and code samples for implementing Amazon Nova Sonic into a workflow. Refer to the User Guide for Amazon Nova or the official Amazon Nova model cookbook GitHub repository.

Observations and next steps

Stats Perform noted technical and operational benefits from leveraging and integrating with AWS. Stats Perform found that the AWS software development kits (AWS SDKs) and code samples enabled rapid integration. The flexible prompt customization and Amazon Nova Sonic Invoke API provided granular control over personalized experiences.

The combination of Amazon Nova FM models, low-latency responses, and the Opta APIs of Stats Perform created a robust system capable of contextual analysis and dynamic UI interactions. This technical foundation, supported by the Accedo Xtend solution, delivered accurate statistical data while maintaining natural conversational engagement.

Looking ahead, Accedo and Stats Perform see an evolution toward voice-first interactions—representing a fundamental shift in user engagement with digital platforms. Key trends include the integration of multimodal context awareness, streamlined immersive platform experiences, and standardized voice interactions across platforms. They observed that with the real-time knowledge integration capabilities and comprehensive features of Amazon Bedrock, with Amazon Nova Sonic, positioned them to lead the next generation of conversational interfaces across immersive, mobile, and traditional computing environments.

Conclusion

The latest generation of XR applications and headsets (powered by AWS and AWS Partners Accedo and Stats Perform) is redefining the sports viewing experience—keeping fans engaged while immersing them in the action with their favorite teams.

We explored how Accedo and Stats Perform leveraged AWS services to power an immersive, data-driven, hands-free viewing experience to keep fans engaged. We also reviewed how the integration with Amazon Nova Sonic FM extended the solution further by enabling speech-to-speech interactions with an AI agent while watching the game.

To learn more about the Accedo Xtend solution contact their Sales Team, and to learn more about Stats Perform and Opta data contact their Sales Team.

Check out more AWS Partners or contact an AWS Representative to know how we can help accelerate your business.

Further reading

Osmar Bento

Osmar Bento

Osmar Bento is a Senior Solution Architect specializing in Direct-to-Consumer experiences for M&E, Gaming, and Sports at AWS. Osmar collaborates with customers to innovate and create tailored solutions using the AWS platform, enhancing their media and entertainment operations.

Emmanuel Soulie

Emmanuel Soulie

Emmanuel Soulie is a Senior Solution Architect focused on Direct-to-Consumer & Streaming solutions for the media, entertainment, gaming and sports industries. Leveraging deep domain expertise, he helps media clients seamlessly adopt cloud technologies to achieve their key business objectives and drive innovation. With a strong commitment to improving user experience and operational efficiency, Emmanuel is dedicated to transforming the digital media landscape through industry-leading AWS solutions.

Jason O'Malley

Jason O'Malley

Jason O’Malley is a Sr. Partner Solutions Architect at AWS supporting partners architecting media, communications, and technology industry solutions. Before joining AWS, Jason spent 13 years in the media and entertainment industry at companies including Conan O’Brien’s Team Coco, WarnerMedia, and Media.Monks. Jason started his career in television production and post-production before building media workloads on AWS. When Jason isn’t creating solutions for partners and customers, he can be found adventuring with his wife and son, or reading about sustainability.