NVIDIA 在 IBC 展示面向广播、体育与全球流媒体的实时 AI

在9月11日至14日于阿姆斯特丹举行的IBC大会上，创意、技术和商业界汇聚一堂，将想法转化为行动，并讨论媒体和娱乐行业的创新。来自170多个国家的44,000多名与会者聚集在14个以上的展厅和户外空间中，探索1,300多场展览，超过600名演讲者分享洞见。

请继续阅读，了解NVIDIA在展会上的重点内容。

媒体公司正日益将AI整合到直播制作、体育、新闻和流媒体工作流程中，以挖掘更丰富的性能洞察、验证视频真实性、增强和本地化内容——所有这些都不会破坏可信的广播环境。

在阿姆斯特丹举行的IBC 2026上，NVIDIA宣布对NVIDIA AI for Media进行重大扩展——这是一个由GPU加速的软件开发工具包（SDK）、NVIDIA NIM微服务、操作指南和蓝图组成的集合，可增强媒体和娱乐工作流程中的音频、视频和增强现实效果——以开创新方式理解动作、验证和增强视频、本地化节目并构建AI驱动的媒体应用程序。

今年早些时候在SIGGRAPH上发布的NVIDIA合成视频检测器（SVD）NIM微服务：https://build.nvidia.com/nvidia/synthetic-video-detector，帮助机构评估素材是真实的还是AI生成的概率，为编辑、内容验证、数字取证和媒体完整性团队在审核过程中提供另一个分析点：https://blogs.nvidia.com/blog/siggraph-news-2026/#synthetic-video。

自最初发布以来，SVD对文本生成视频内容的准确率已达到99.3%，对图像生成视频内容的准确率达到97.7%，在难度较大的图像生成视频案例上尤其进步显著。

Dalet正在将SVD整合到面向新闻机构的安全云托管验证工作流程中。这允许编辑团队通过SVD提交素材，在Dalet界面内检查并审核生成的评分和元数据。

TwelveLabs 宣布其首个基于公司视频智能平台构建的应用程序——Compliance by TwelveLabs 正式推出，该应用帮助媒体和广播团队快速对内容进行地区和自定义合规标准的筛查。该解决方案集成了 SVD，以添加帧级真实性信号和置信度评分，使媒体团队能够在同一合规工作流程中识别潜在的合成媒体。

Wowza，其 Wowza Streaming Engine（https://www.wowza.com/streaming-engine）媒体服务器技术支持超过 170 个国家的 35,000 多个视频部署，将通过 Wowza Video Intelligence Framework（https://www.wowza.com/video-intelligence-framework）分发 SVD。该解决方案由 NVIDIA 加速基础设施驱动，使广播公司、流媒体提供商和其他组织能够实时分析直播视频流并提取检测到的对象、场景及 AI 生成痕迹的数据。它可以在本地、边缘计算、云端、混合部署或完全隔离环境中部署和运行，为组织提供对关键媒体工作流程的更大控制权。

NVIDIA 3D Body Pose 可以从单摄像机捕获的视频中估算二维和三维人体关节位置及角度，帮助在无需基于标记的捕获系统的情况下将动作转化为结构化数据。

对于体育组织，这些数据可以支持球员和运动员动作跟踪、生物力学和性能分析、回放增强、裁判判决流程、球员安全应用以及虚拟互动和沉浸式体验。

这项技术还可以为内容创作工作流提供结构化的人体动作数据。当映射到兼容的角色骨架时，关节和动作数据可作为动画阻挡、数字替身、角色重定向和虚拟制作体验的输入。

Vizrt 正在实时虚拟演播室环境中使用 Body Pose 技术，跟踪的身体动作驱动实时 3D 灯光效果，如反射、阴影和环境渲染。

视频帧生成 (VFG) 通过使用生成式人工智能在视频的原始帧之间创建新帧，使视频运动看起来更加平滑。它可以将帧率提高 2 倍或 4 倍，同时保持视觉质量和时间一致性，从而实现更流畅的体育赛事、慢动作重播、现场媒体和其他高动感视频体验。VFG 还支持生成式 AI 视频工作流中的帧率转换和帧提升。

Ross Video 正在将 VFG 集成到其 Rio Replay 平台中，以为体育制作创建 AI 辅助的慢动作视频。

该工作支持体育重播的 6 倍慢动作生成。开发工作正在进行，以实现 8 倍插帧，这意味着生成的中间帧可以为重播团队提供更平滑的运动，而不需要由超高帧率源摄像机捕捉每一帧。

NVIDIA 视频超分辨率 (VSR) 使用 AI 对视频进行升级，同时减少噪点、模糊和压缩伪影。新的流式模式让开发者可以在实时性能和更高图像质量之间进行选择，而可调控的设置有助于实现所需的增强效果。VSR 还增加了 10 位视频支持，并提升了整体性能和质量。VSR 可通过 NVIDIA 视频特效 SDK 和用于流媒体、广播、会议、视频播放和内容创作应用的 NIM 微服务获得。

该技术可以通过统一接口支持视频播放器、会议应用、创作工具、流媒体服务、转码器和广播系统。

NVIDIA TrueHDR 可将标准动态范围（SDR）视频实时转换为高动态范围（HDR）输出，亮度最高可达约 2,000 尼特，同时保持局部对比度并根据内容调节亮度。

VSR、VFG 和 TrueHDR 可以在单一视频特效管道中结合使用——帮助媒体公司增强现有内容库，用于流媒体、转码、游戏和创作者工作流。

NVIDIA LipSync：https://build.nvidia.com/nvidia/lipsync 和主动发言者检测 (Active Speaker Detection)：https://build.nvidia.com/nvidia/active-speaker-detection NIM 微服务帮助开发者为采访、新闻、体育、娱乐及其他可能在屏幕上出现多人的节目构建本地化系统。

LipSync 将输入视频中的口型动作转换为与目标音频轨道相匹配，同时保留自然的头部姿势、眨眼和身体动作。新版本改进了面部遮挡处理，并更好地保留牙齿、嘴唇和面部纹理。

新的主动说话者检测 NIM 微服务不再需要对多个音频轨道进行说话者分离，新增语音活动检测，并通过 gRPC 接口扩展了 NIM 微服务部署支持以及更广泛的 GPU 兼容性。

NDI：https://ndi.video/stories/press/ndi-nvidia-ai-multilingual-content/ 使用 NVIDIA AI for Media，包括 NVIDIA LipSync NIM 微服务，实现现有广播工作流程中的实时翻译、口型同步配音和区域语言适配。通过从同一媒体流生成多语言体验，这种方法可以帮助广播公司触及全球观众，同时减少传统多语言分发所需的带宽、基础设施和制作复杂性。

Studio Voice 包含基于 NVIDIA Studio Voice NIM 微服务的新麦克风配置文件，使用户对增强语音的音色特性有更多控制。

该功能旨在抑制背景噪声、减少房间混响并提高语音清晰度，然后将增强后的输出塑造成所选的麦克风配置文件，以实现更专业的实时交流、流媒体、播客和内容创作。

试用 NVIDIA AI for Media NIM 微服务：https://build.nvidia.com/models?label=nvidia+ai+for+media。查看 IBC 2026 的最新 NVIDIA 及合作伙伴工作流程：https://www.nvidia.com/en-us/events/ibc/。

随着广播公司、流媒体服务和体育组织采用软件和 AI，直播内容背后的基础设施正变得更加灵活、更互联，并越来越多地建立在共享加速计算之上。

Media Exchange Layer (MXL) 与 NVIDIA Holoscan for Media 的集成：https://developer.nvidia.com/holoscan-for-media 加速了这一过渡——提供了帮助媒体应用连接并协同运行的通用交换层。

Holoscan for Media 是一个开放参考架构和开发者工具包，用于构建面向 AI 的媒体功能和软件定义实时制作的应用程序。MXL 为这些基于软件的媒体功能提供了一种开放方式，以在分布式环境中交换实时视频、音频和数据。

随着生产功能向软件化转移，开发者可以构建共享加速基础设施、动态连接并独立演进的应用程序。这可以帮助媒体公司更高效地使用基础设施，更快地引入新功能，并减少应用程序之间所需的自定义集成工作量。

这种整合也为实时媒体中的 AI 打下了更坚实的基础。AI 处理、视频应用和传统媒体功能可以越来越多地在相同的加速基础设施和相同的软件定义环境中运行。

对于技术供应商而言，这扩大了构建能够跨更广泛、多供应商生态系统运行的应用程序的机会。对于媒体公司而言，这为基础设施提供了一条可随格式、应用程序和 AI 功能变化而适应的路径。

在 IBC 的 EBU 展位 10.D21 观看演示：https://directory.ibc.org/8_0/floorplan/?hallID=F&level=1&st=exhibitor&selectedBooth=booth%7E10.D21。了解更多关于 Holoscan for Media 的信息：https://developer.nvidia.com/holoscan-for-media。

体育正在成为 AI 更广泛转变的试验场：从通用模型向基于专有数据构建的微调开放模型发展。

NVIDIA Sports Intelligence Playbooks：https://animated-adventure-8ggoqky.pages.github.io/latest/ 旨在加速这一转变。它们为联盟、媒体公司和技术提供商提供了结构化框架，以在自己的体育录像和注释上微调 NVIDIA 开放模型，从而创建能够理解比赛规则、球员、得分、策略和体育独特背景的多模态 AI。

体育组织拥有大量难以被竞争者复制的专有视频、元数据和绩效信息。Playbooks 提供了一个实用蓝图，将这些资产转化为 AI 能力，从而支持新的分析产品、媒体体验、自动化工具和收入来源。

这些操作手册涵盖了 AI 生命周期的各个阶段，包括数据准备、微调、推理、评估、优化和部署，并整合了 NVIDIA 的技术，包括 Nemotron：https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/ 、NeMo AutoModel：https://docs.nvidia.com/nemo/automodel 、Megatron Bridge：https://docs.nvidia.com/nemo/megatron-bridge/latest/ 、NIM 微服务：https://www.nvidia.com/en-us/ai-data-science/products/nim-microservices/ 以及 NVIDIA 加速计算：https://www.nvidia.com/en-us/data-center/solutions/accelerated-computing/。

通过提供从模型定制到生产的集成路径，Sports Intelligence Playbooks 可以降低构建专业体育 AI 的成本和复杂性，同时提升其在计算、软件和推理堆栈上的需求。

早期测试展示了领域专业化的潜力。在使用与训练中类似的问题形式评估之前未见过的素材时，多项选择题准确率从约 53% 提升至 94%，开放式评估从约 5.7% 提升至 66%。

Machina Sports 正在将 Sports Intelligence Playbooks 与其体育本地数据、评估和代理基础设施整合，使权益持有者能够将专有媒体和专业知识转化为可在实时制作、内容和粉丝体验中部署的私人智能。

随着代理式 AI 的日益采用，这一机会也在扩大。借助 NVIDIA AI-Q Blueprint：https://build.nvidia.com/nvidia/aiq，组织可以将其特定领域的体育模型用作代理中的专家智能，从而在视频、企业数据和软件系统中进行推理，将操作手册从体育理解扩展到决策和自动化。

Wowza 正在将视觉语言模型（包括 NVIDIA Cosmos 3 和 Nemotron）整合到 Wowza 视频智能框架中，通过 NVIDIA Sports Intelligence Playbooks 进行微调，以检测直播流中的体育特定瞬间并缩短行动时间。

探索 NVIDIA 体育智能剧本：https://github.com/NVIDIA/sports-intelligence-playbooks。

通过直播节目触达全球观众不仅仅是翻译文字。语言的细微差别、语音、时间把控、面部动作、字幕和屏幕上的图形必须实时协同工作，同时保持原节目内容的编辑意图和制作质量。

为了帮助广播公司、体育联盟、版权持有者和流媒体服务将这些元素整合到统一的软件定义实时本地化工作流程中，NVIDIA 正将其内容本地化技术引入 NVIDIA Holoscan for Media 开发者工具包。该参考工作流程专为广播和流媒体开发者设计，可实现字幕、翻译音频、配音、同步视频和本地化图形。

Holoscan for Media 的内容本地化为本地化技术如何在软件定义的广播应用中协同工作提供了参考。开发者可以根据每个节目、市场或发行渠道选择所需的功能，而无需为每个本地化版本部署独立基础设施。

Holoscan for Media 的内容本地化整合了 NVIDIA AI for Media 的最新进展，包括在面部部分被遮挡时改进的唇动同步（LipSync）以及增强的主动发言者检测（Active Speaker Detection），帮助应用识别多人物场景中的发言者。

本地化可以改变直播节目覆盖范围和经济效益。一个共享的、可组合的工作流程可以帮助媒体公司更快地推出地区性报道，服务更多观众，并为个别市场定制体验，同时保持直播制作所需的时间掌控、视觉环境和编辑控制。

来自 AI-Media、CAMB.AI、Chyron 和 Panjaya 的技术各自解决了 Holoscan for Media 内容本地化中的特定环节，从调整语音和屏幕呈现，到创建多语言字幕和翻译音频、实现图形本地化，以及在直播和点播内容中保持表达和身份的一致性。

内容本地化技术还支持基于文件、流媒体和后期制作的应用。开发人员可以使用应用程序编程接口进行按需工作流程，并在本地化必须作为实时媒体环境的一部分运行时使用 Holoscan for Media 参考工作流。它们共同为跨生产和分发构建多语言媒体服务提供了一致的基础。

了解更多关于 NVIDIA Holoscan for Media 的信息：https://developer.nvidia.com/holoscan-for-media 以及关于媒体的 AI 信息：https://developer.nvidia.com/topics/ai/generative-ai/ai-for-media。

At the IBC conference, running Sept. 11-14 in Amsterdam, the creative, technology and business communities are coming together to turn ideas into action and discuss innovations across the media and entertainment industries. More than 44,000 attendees from 170+ countries are gathering to explore 1,300+ exhibitions in 14+ halls and outdoor spaces, with over 600 speakers delivering insights.

Read on to learn more about what NVIDIA’s highlighting at the show.

Media companies are increasingly integrating AI into live production, sports, news and streaming workflows to unlock richer performance insights, verify video authenticity, and enhance and localize content — all without disrupting trusted broadcast environments.

At IBC 2026 in Amsterdam, NVIDIA announced a major expansion to NVIDIA AI for Media — a collection of GPU-accelerated software development kits (SDKs), NVIDIA NIM microservices, playbooks, and blueprints that enhance audio, video and augmented-reality effects for media and entertainment workflows — to unlock new ways to understand motion, verify and enhance video, localize programming and build AI-powered media applications.

The NVIDIA Synthetic Video Detector ：https://build.nvidia.com/nvidia/synthetic-video-detector (SVD) NIM microservice, announced earlier this year at SIGGRAPH ：https://blogs.nvidia.com/blog/siggraph-news-2026/#synthetic-video , helps organizations assess the probability of whether footage is authentic or AI-generated, giving editorial, content-authentication, digital-forensics and media-integrity teams another point of analysis in their review process.

Since its initial release, SVD’s accuracy has reached 99.3% for text-to-video content and 97.7% for image-to-video content, with especially large gains on difficult image-to-video cases.

Dalet is integrating SVD into a secure, cloud-hosted verification workflow for news organizations. This allows editorial teams to submit footage through SVD, inspect and review the resulting scores and metadata within a Dalet interface.

TwelveLabs announced the general availability of Compliance by TwelveLabs, its first application built on the company’s video intelligence platform, helping media and broadcast teams rapidly screen content against regional and custom compliance standards. The solution integrates SVD to add frame-level authenticity signals and confidence scores, enabling media teams to identify potentially synthetic media within the same compliance workflow.

Wowza , whose Wowza Streaming Engine ：https://www.wowza.com/streaming-engine media server technology powers more than 35,000 video deployments across over 170 countries, will distribute SVD through the Wowza Video Intelligence Framework ：https://www.wowza.com/video-intelligence-framework . The solution, powered by NVIDIA-accelerated infrastructure, will enable broadcasters, streaming providers and other organizations to analyze live video feeds and extract data around detected objects, scenes and signs of AI generation in real time. It can be deployed and run on premises, at the edge, in the cloud, across hybrid deployments or fully air-gapped, giving organizations greater control over critical media workflows.

NVIDIA 3D Body Pose estimates 2D and 3D human joint locations and angles from video captured by a single camera, helping turn motion into structured data without marker-based capture systems.

For sports organizations, that data can support player and athlete movement tracking, biomechanics and performance analysis, replay enhancement, officiating and adjudication workflows, player-safety applications, and virtual interaction and immersive experiences.

The technology can also provide structured human-motion data for content-creation workflows. When mapped to a compatible character rig, joint and motion data can serve as input for animation blocking, digital doubles, character retargeting and virtual-production experiences.

Vizrt is using Body Pose technology in live virtual-studio environments, with tracked body movement driving real-time 3D lighting effects such as reflections, shadows and environmental rendering.

Video Frame Generation (VFG) makes video motion appear smoother by using generative AI to create new frames between the original frames of a video. It can increase frame rates by 2x or 4x while preserving visual quality and temporal consistency, enabling more fluid sports, slow-motion replays, live media and other high-motion video experiences. VFG also supports frame-rate conversion and frame boosting for generative AI video workflows.

Ross Video is integrating VFG into its Rio Replay platform to create AI-assisted slow-motion video for sports production.

The work supports 6x slow-motion generation for sports replay. Development is underway toward 8x interpolation, meaning generated intermediate frames can give replay teams smoother motion without requiring every frame to be captured by an ultrahigh-frame-rate source camera.

NVIDIA Video Super Resolution (VSR) uses AI to upscale video while reducing noise, blur and compression artifacts. New streaming modes let developers choose between real-time performance and higher image quality, while adjustable controls help achieve the desired level of enhancement. VSR also adds 10-bit video support and improves overall performance and quality. VSR is available through the NVIDIA Video Effects SDK and a NIM microservice for use in streaming, broadcast, conferencing, video playback and content-creation applications.

The technology can support video players, conferencing applications, creator tools, streaming services, transcoders and broadcast systems through a common interface.

NVIDIA TrueHDR converts standard-dynamic-range video into high-dynamic-range output in real time, reaching up to approximately 2,000 nits while preserving local contrast and adapting brightness to the content.

VSR, VFG and TrueHDR can be combined within a single video-effects pipeline — helping media companies enhance existing content libraries for streaming, transcoding, gaming and creator workflows.

The NVIDIA LipSync ：https://build.nvidia.com/nvidia/lipsync and Active Speaker Detection ：https://build.nvidia.com/nvidia/active-speaker-detection NIM microservices help developers build localization systems for interviews, news, sports, entertainment and other programming where multiple people may appear on screen.

LipSync transforms mouth movement in an input video to match a target audio track while preserving natural head pose, blinking and body movement. The new release improves facial occlusion handling and better preserves teeth, lip and facial textures.

The new Active Speaker Detection NIM microservice no longer requires speaker diarization for multiple audio tracks, adds voice activity detection and expands NIM microservice deployment support through a gRPC interface and broader GPU compatibility.

NDI ：https://ndi.video/stories/press/ndi-nvidia-ai-multilingual-content/ is using NVIDIA AI for Media, including the NVIDIA LipSync NIM microservice, to enable real-time translation, lip-synced dubbing and regional language adaptation within existing broadcast workflows. By generating multiple language experiences from a common media stream, the approach can help broadcasters reach global audiences while reducing the bandwidth, infrastructure and production complexity traditionally required for multilingual distribution.

Studio Voice includes new Microphone Profiles built on NVIDIA Studio Voice NIM microservices, giving users more control over the tonal character of enhanced speech.

The capability is designed to suppress background noise, reduce room reverberation and improve speech clarity, then shape the enhanced output into a selected microphone profile for more polished live communications, streaming, podcasting and content creation.

Try NVIDIA AI for Media NIM microservices ：https://build.nvidia.com/models?label=nvidia+ai+for+media . See the latest NVIDIA and partner workflows at IBC 2026 ：https://www.nvidia.com/en-us/events/ibc/ .

As broadcasters, streaming services and sports organizations adopt software and AI, the infrastructure behind live content is becoming more flexible, more connected and increasingly built on shared accelerated computing.

The integration of Media Exchange Layer (MXL) with NVIDIA Holoscan for Media ：https://developer.nvidia.com/holoscan-for-media accelerates that transition — providing the common exchange layer that helps media applications connect and operate together.

Holoscan for Media is an open reference architecture and developer toolkit for building AI-powered media functions and applications for software-defined live production. MXL adds an open way for those software-based media functions to exchange live video, audio and data across a distributed environment.

As production functions move into software, developers can build applications that share accelerated infrastructure, connect dynamically and evolve independently. That can help media companies use infrastructure more efficiently, introduce new capabilities faster and reduce the amount of custom integration required between applications.

The integration also creates a stronger foundation for AI in live media. AI processing, video applications and traditional media functions can increasingly operate on the same accelerated infrastructure and in the same software-defined environment.

For technology vendors, this expands the opportunity to build applications that can work across broader, multi-vendor ecosystems. For media companies, it creates a path toward infrastructure that can adapt as formats, applications and AI capabilities change.

See the demo at IBC in EBU Stand 10.D21 ：https://directory.ibc.org/8_0/floorplan/?hallID=F&level=1&st=exhibitor&selectedBooth=booth%7E10.D21 . ​Learn more about Holoscan for Media ：https://developer.nvidia.com/holoscan-for-media .

Sports is becoming a proving ground for a broader shift in AI: from general-purpose models toward fine-tuned open models built on proprietary data.

NVIDIA Sports Intelligence Playbooks ：https://animated-adventure-8ggoqky.pages.github.io/latest/ are designed to accelerate that transition. They give leagues, media companies and technology providers structured frameworks to fine-tune NVIDIA open models on their own sports footage and annotations, creating multimodal AI that can understand the rules, players, scoring, strategy and context unique to a sport.

Sports organizations hold large volumes of proprietary video, metadata and performance information that are difficult for competitors to replicate. The playbooks provide a practical blueprint for converting those assets into AI capabilities that can underpin new analytics products, media experiences, automation tools and revenue streams.

The playbooks span the AI lifecycle, including data preparation, fine-tuning, inference, evaluation, optimization and deployment, and bring together NVIDIA technologies including Nemotron ：https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/ , NeMo AutoModel ：https://docs.nvidia.com/nemo/automodel , Megatron Bridge ：https://docs.nvidia.com/nemo/megatron-bridge/latest/ , NIM microservices ：https://www.nvidia.com/en-us/ai-data-science/products/nim-microservices/ and NVIDIA accelerated computing ：https://www.nvidia.com/en-us/data-center/solutions/accelerated-computing/ .

By providing an integrated path from model customization to production, Sports Intelligence Playbooks can reduce the cost and complexity of building specialized sports AI while increasing demand across its compute, software and inference stack.

Early testing demonstrates the potential of domain specialization. When evaluated on previously unseen footage using question formats similar to those used in training, multiple-choice accuracy increased from approximately 53% to 94% and open-ended evaluation from approximately 5.7% to 66%.

Machina Sports is integrating Sports Intelligence Playbooks with its sports-native data, evaluation and agent infrastructure, enabling rights holders to turn proprietary media and expertise into private, deployable intelligence for live production, content and fan experiences.

The opportunity also expands as agentic AI becomes increasingly adopted. With the NVIDIA AI-Q Blueprint ：https://build.nvidia.com/nvidia/aiq , organizations can use their domain-specific sports models as expert intelligence within agents that reason across video, enterprise data and software systems, extending the playbook from sports understanding into decision-making and automation.

Wowza is integrating vision language models, including NVIDIA Cosmos 3 and Nemotron, into the Wowza Video Intelligence Framework, fine-tuned through NVIDIA Sports Intelligence Playbooks to detect sports-specific moments in live streams and reduce time to action.

Explore NVIDIA Sports Intelligence Playbooks ：https://github.com/NVIDIA/sports-intelligence-playbooks .

Reaching global audiences with live programming requires more than translating words. Language nuances, voice, timing, facial movement, captions and onscreen graphics must work together in real time, while preserving the editorial intent and production quality of the original program.

To help broadcasters, sports leagues, rights holders and streaming services bring these elements into a unified, software-defined, real-time localization workflow, NVIDIA is bringing its Content Localization technologies to the NVIDIA Holoscan for Media developer toolkit. Designed for broadcast and streaming developers, the reference workflow enables captions, translated audio, dubbing, synchronized video and localized graphics.

Content Localization with Holoscan for Media provides a reference for how localization technologies can work together in software-defined broadcast applications. Developers can select the capabilities needed for each program, market or distribution channel rather than deploying separate infrastructure for every localized version.

Content Localization with Holoscan for Media incorporates the latest advancements from NVIDIA AI for Media, including improved LipSync when faces are partially obscured and enhanced Active Speaker Detection to help applications identify who’s speaking in multi-person scenes.

Localization can transform the reach and economics of live programming. A shared, composable workflow can help media companies introduce regional coverage faster, serve more audiences and tailor experiences for individual markets — while preserving the timing, visual context and editorial control required for live production.

Technologies from AI-Media , CAMB.AI , Chyron and Panjaya each address a specific part of content localization with Holoscan for Media, from adapting voice and onscreen delivery to creating multilingual captions and translated audio, localizing graphics, and preserving expression and identity across live and on-demand content.

The Content Localization technologies also support file-based, streaming and post-production applications. Developers can use application programming interfaces for on-demand workflows and the Holoscan for Media reference workflow when localization must run as part of a live media environment. Together, they provide a consistent foundation for building multilingual media services across production and distribution.

Learn more about NVIDIA Holoscan for Media ：https://developer.nvidia.com/holoscan-for-media and AI for Media ：https://developer.nvidia.com/topics/ai/generative-ai/ai-for-media .