AWS for M&E Blog

Warner Bros. Discovery Sports engages fans using scaled audio data analysis on AWS

This blog is co-authored by Steve Cockett, Sports Data Adviser – Tech, at Warner Bros. Discovery and Andrea Fanfani, Principal Product Manager – Tech, at Warner Bros. Discovery.

Feed latency and marker accuracy are two critical challenges in live sports broadcasting. Warner Bros. Discovery (Nasdaq: WBD) has developed a solution to tackle both as it continues to revolutionize sports streaming. It has implemented an innovative audio analysis solution to enhance data accuracy and the fan experience on its streaming platforms.

Now dubbed Metronome, the solution uses Amazon Web Services (AWS) serverless architecture to analyze audio spectrograms. It can detect key moments (such as kick offs) through referee whistles and validate significant events through crowd excitement levels. This automated solution improves the precision of timeline markers for soccer matches, and also increases operational efficiency through automated alerts.

Timeline markers are an important feature for sports fans, allowing them to instantly access key moments in matches without tediously searching through video. This is particularly valuable for soccer fans who want to revisit goals, cards, or other crucial moments, making the viewing experience more interactive. Since fans increasingly expect sophisticated viewing options that put them in control, these markers transform passive watching into an active, engaging experience where viewers never miss important moments.

In a proof-of-concept, Metronome proved particularly effective, with a 96% accuracy rate in a test match, successfully detecting 24 out of 25 match events within a 1000 millisecond (ms) threshold. By leveraging AWS services (such as AWS Lambda, AWS Step Functions, Amazon DynamoDB, and Amazon Rekognition) WBD created a cost-effective, near real-time solution that significantly enhances the viewing experience for their customers.

We will explain how WBD has built on the success of the original proof-of-concept, scaling to production, and leveraging the crowd excitement algorithm to enhance the fan experience in new and innovative ways.

Scaling to production

Metronome has been deployed into a production environment to improve real-time monitoring for the live operations teams. It is deployed across all of its live soccer assets using data driven timeline markers and alerts. It is enabled for HBO Max and discovery+ users, covering different kinds of game competitions. For example, for TNT Sports football coverage, which streams on discovery+ in the UK, including the Premier League, UEFA Champions League, Europa League and Conference League.

From the 2025-2026 season, TNT Sports UK expanded football offer will include premium coverage of more FA Cup matches, the world’s oldest cup competition, than ever before. Plus, six new competitions will be shown on TNT Sports in selected Latin American markets, for which Metronome will be used.

Currently, Metronome provides the WBD operation teams with real-time data on two key areas to help increase the accuracy of end user timeline markers:

  • Kick off accuracy
  • Timeline marker accuracy

Kick off accuracy

Metronome generates an independent calculation of the kick off time based upon the referee’s whistle or the TV clock graphic in the video feed.  This provides the operations team an instant comparison with the video sync marker data. Any significant differences can be manually investigated immediately and corrected, if required.

A screenshot from the Metronome UI with two thumbnails side by side. It shows the original frame with kickoff timing (20:01:40.300) compared to the corrected frame with the Metronome timing (20:01:43.150).

Figure 1: Metronome comparing kick off times.

Timeline marker accuracy

Throughout the match, operators can access video clip comparisons of each key timeline marker, with one clip based on the sports data provider and another clip generated using the corrected timestamp from the Metronome audio analysis.

A screenshot from the Metronome UI with two thumbnails side by side. It shows the original frame with goal timing (21:27:09.914) compared to the corrected frame with the Metronome timing (21:27:12.660).

Figure 2: Metronome comparing shot on target timing.

This helps the live operations team in two main ways:

  • Any misalignment issues between the timeline marker and target action can be checked immediately within the video player to triage the likely cause:
    1. Issue with the client video player
    2. Timestamp error from the sports data provider
    3. Timestamp error from the sports data proxy API serving the WBD streaming platforms
  • A timeline marker accuracy table helps operators to notice any patterns in the video and data synchronization for a given match. They can then communicate these to the WBD sports data provider, if required.
A table of Timeline Marker Accuracy results showing each event in the match (goals, shorts on target, and so on) along with their excitement score and timing offset.

Figure 3: Table showing timeline marker accuracy and excitement score.

Scaling through serverless architecture

The ability to rapidly scale across multiple competitions and territories is made possible through AWS serverless architecture. By leveraging services (such as AWS Lambda, Amazon DynamoDB, and Amazon EventBridge), Metronome automatically adjusts computing resources based on the number of concurrent matches being processed.

An architecture diagram showing the serverless architecture components of the workflow.

Figure 4: Metronome serverless architecture.

During peak periods, the serverless infrastructure automatically provisions the necessary resources to process audio streams and generate markers in real-time. This eliminates the need to maintain dedicated infrastructure that would otherwise remain underutilized during periods of lower activity.

The serverless approach provides several key benefits:

  • Dynamic resource allocation: Compute resources automatically scale up or down based on the number of active matches
  • Cost optimization: WBD only pays for the actual processing time used during live events
  • Operational efficiency: No need to manage or maintain underlying infrastructure
  • High availability: Automatic failovers and redundancies verify reliable operations across different AWS availability zones

This architecture provides WBD the ability to process hundreds of hours of live sports content weekly across multiple competitions, while maintaining consistent performance and reliability. As the WBD sports portfolio continues to grow, with upcoming additions, the serverless infrastructure will seamlessly scale to meet the increased demand without requiring significant architectural changes.

The ability of Metronome to handle rapid scaling, while maintaining cost efficiency, demonstrates the power of cloud-native architectures in modern sports broadcasting. This approach helps WBD to focus on enhancing the viewer experience rather than managing infrastructure. It facilitates rapid innovation and feature development across their growing sports portfolio.

The deployment and scaling to production of the solution has been a success and is recognized by senior stakeholders. “Metronome gives us another trusted source to validate the accuracy of our soccer sport data for all key moments. This allows our live operations team the best chance to resolve any video and data synchronization issues as soon as possible. It ensures our timeline markers provide the best experience to our users across both HBO Max and discovery+ platforms,” said Andrea Fanfani, Senior Director of Product Management at WBD.

Future product advancements of Metronome

WBD product teams are developing feature enhancements that both expand capabilities and leverage data insights across broader applications. The ability of Metronome to automatically detect significant moments through audio analysis opens new possibilities for automated highlight generation, personalized content delivery, and enhanced fan engagement across the WBD growing sports portfolio.

Automating kick off sync marker

For timeline markers to be shown within the video player, a kick off sync marker is required for a given match to synchronize the video and data streams. Whilst Metronome currently detects kick off events, it is only used for validation. An operator is still ultimately responsible for setting the kick off time.

On some match-days there can be up to 16 concurrent soccer matches being played, which can lead to short delays in the kick off marker being set manually by the operations team. WBD is planning to enable the kick off detection of Metronome to automatically set this sync marker for all matches, provided that no marker has already been created by the operations team at the time it is calculated. This will lower the workload for operators on match-days where they are monitoring multiple games. They will be able to review the sync marker automatically calculated and, if required, make a small adjustment.

Timeline marker promotion for shots off target

Within the WBD enhanced streaming experience on HBO Max and discovery+, users are presented with the best moments of the match within its timeline feature. A user can view any moment, at any time, elements such as goals, cards, penalties, shots on target, and the start and end of a period.

A screenshot of the TNT streaming platform showing a soccer match with timeline markers for match events running along the bottom of the screen.

Figure 5: End user application with timeline markers.

Due to the high number of Shots Off Target in each match, and their varied excitement level, they are not displayed on the timeline. However, within many matches, some of the highest excitement moments are these types of events. For example, a miss on an open goal or an amazing shot that just missed the target.

Prior to Metronome there was no way to identify these crucial moments automatically through sports data. Now, with the crowd excitement feature of Metronome, WBD has an excitement rating for each marker.

A screenshot from the Metronome UI with two thumbnails side by side. It shows the original frame with shot off target timing (20:17:27.300) compared to the corrected frame with the Metronome timing (20:17:29.573).

Figure 6: Screen capture of Metronome comparing shot off target timing.

Crowd excitement graph

Metronome produces a crowd excitement graph, based on the crowd excitement algorithm described in How Warner Bros. Discovery uses audio analysis to improve data accuracy and enrich the fan experience. The graphs help operators to visualize each timeline marker and corresponding crowd excitement. Using a variable threshold, operators can filter for the most exciting moments in the match.

A crowd excitement line graph from Metronome highlighting the peak excitement points and how they correlate to in match markers.

Figure 7: Crowd excitement visualization.

Using the UEFA Europa League Final between Tottenham Hotspur and Manchester United for example, the highest crowd excitement rating, for a shot on or off target, throughout the match was a shot off target by Amad Diallo. It was at the 16-minute mark, with a raw data excitement score of 50.2.

WBD is now working towards using crowd excitement ratings to include markers with high crowd excitement above a predetermined threshold. These new markers would be shown on the timeline so users can always re-watch the best moments of every match.

Crowd excitement for quick recap highlights

In addition, WBD is exploring using crowd excitement data to help with the creation of quick recap highlights. The highlights would be available to users, prior to joining the live stream, while a match is in-progress.

These highlights would combine mandatory marker types (such as goals and missed penalties). However, in the instance of these not being available, crowd excitement ratings would be used to provide users with the next best moments of the match so far.

Using the example of the UEFA Europa League Final between Tottenham Hotspur and Manchester United, which finished 1-0, the top three highlight moments of the match detected were:

  1. 16’ Amad Diallo shot off target (with the highest crowd excitement score)
  2. 42’ Brennan Johnson goal—Tottenham 1-0 Man Utd (a current mandatory marker type)
  3. 68’ Rasmus Højlund shot on target (with the second highest crowd excitement score)

Conclusion

The continuing evolution Metronome demonstrates how AWS serverless architecture helps media organizations to rapidly innovate and scale their operations. What began as a focused effort to improve timeline marker accuracy has grown into a comprehensive system supporting multiple competitions across the global WBD sports portfolio. The solution now enhances the viewing experience for fans watching premier sporting events on the WBD platforms.

Through AWS managed services, WBD has transformed its operational capabilities across multiple dimensions. The team can now deploy complex audio analysis capabilities without managing infrastructure, while seamlessly scaling operations across territories and competitions. The Metronome platform enables rapid innovation with new features like marker event promotion, all while improving operational efficiency and maintaining broadcast-grade reliability.

As WBD continues to expand its premium sports rights portfolio and further enhance its streaming platforms, the flexibility and scalability of AWS services provide a foundation for future innovation. The success of this project showcases how cloud-native architectures can transform traditional broadcast operations. Media companies can focus on creating compelling viewer experiences rather than managing technical infrastructure. This demonstrates how the comprehensive suite of AWS services helps media organizations to stay at the forefront of innovation in sports broadcasting.

Contact an AWS Representative to know how we can help accelerate your business.

Further reading

Andrew Lee

Andrew Lee

Andrew is a Senior Media Cloud Architect with AWS Professional Services based in Amsterdam, The Netherlands. In his spare time he loves playing guitar, reading a good book or traveling around the world.

Preeti Devulapalli

Preeti Devulapalli

Preeti Devulapalli is an Associate Data and AI/ML Engineer with AWS Professional Services, specialized in building generative AI applications (including RAG and agentic AI use cases).

Vivek Thacker

Vivek Thacker

Vivek is a Senior Engagement Manager with AWS Professional Services supporting Global clients across Telecom, Media & Entertainment and Sports verticals. He specializes in leading large-scale cloud migration and digital transformation engagements that drive strategic business outcomes and accelerate customers' cloud adoption journey.