How To Generate Multilingual Sports Commentary at Scale

A football match draws 400 million viewers across six continents. The original commentary is in English. Fans in Brazil, Japan, Saudi Arabia, and France hear nothing in their own language, or they hear a flat, emotionless voiceover that strips out every ounce of excitement from the broadcast.

Traditional dubbing for live sports does not exist at scale. Hiring commentators for every language, for every match, across an entire season is financially and logistically impossible for most broadcasters. Pre-recorded commentary cannot keep pace with live events.

Multilingual sports commentary at scale requires a different approach: AI-powered live dubbing that clones the original commentator's voice, preserves emotion, and delivers the output in real time across 150+ languages.

Here is how to build that workflow, step by step.

What Is Multilingual Sports Commentary?

Multilingual sports commentary is the process of producing play-by-play and color commentary in multiple languages simultaneously during a live or recorded sports event. The goal is to give every fan the same quality of experience, regardless of the language they speak.

Historically, broadcasters achieved multilingual commentary by hiring separate commentary teams for each target language. A single Premier League match might need English, Spanish, Mandarin, Arabic, and Portuguese commentary teams, each working from a dedicated broadcast booth. The cost per language per match runs into thousands of dollars, and scaling beyond five or six languages becomes impractical.

AI-powered live dubbing changes this equation. A single source commentary feed can now be dubbed into dozens of languages in real time, with the original commentator's vocal identity and emotional delivery preserved in every version.

Step 1: Prepare Your Source Commentary Feed

Every multilingual commentary workflow starts with a clean source audio feed. The quality of your output depends directly on the quality of your input.

Ensure your source commentary uses a dedicated audio track, separated from crowd noise, stadium effects, and music. Most professional broadcasts already produce an isolated commentary feed (known as an international sound mix plus commentary). If your feed combines commentary with ambient audio, use speaker diarization to automatically identify and separate individual speakers from background noise before processing.

The cleaner the source, the more accurate the transcription and the more natural the dubbed output.

Step 2: Select Your Target Languages

Not every broadcast needs 150+ languages. Start by identifying where your audience actually is.

Review your viewership data by region. A NASCAR broadcast might prioritize Spanish, Portuguese, and French. A cricket match on FanCode might need Hindi, Tamil, Telugu, and Bengali. A Ligue 1 match might target Italian, English, Arabic, and Japanese.

Prioritize languages based on three factors:

• Audience size in each language market
• Revenue potential from regional broadcast rights and sponsorships
• Fan engagement metrics from social media and digital platforms

Once you have your priority list, you can scale to additional languages incrementally without reworking your entire pipeline.

Step 3: Set Up Real-Time Transcription

Live commentary needs to be transcribed into text before translation and dubbing can happen. Real-time speech-to-text processing converts the spoken commentary into a text stream with minimal latency.

The transcription layer needs to handle sports-specific vocabulary: player names, team names, stadium names, tactical terminology, and colloquial phrases that commentators use in the heat of the moment. A dictionary feature allows you to define pronunciation and terminology rules for proper nouns and domain-specific terms, ensuring "Mbappé" is not transcribed as "Bappay" and "hat trick" is not misinterpreted.

Accurate transcription is the foundation. Errors at this stage compound through every downstream step.

Step 4: Translate Commentary With Context Awareness

Translation for sports commentary is not the same as translating a document. Commentary is fast, emotional, and full of cultural references. A direct word-for-word translation sounds robotic and loses the energy that makes sports broadcasting compelling.

Context-aware AI translation analyzes the tone, pace, and intent of each phrase before producing the target language output. When a commentator shouts "What a strike!" the translation needs to carry the same level of excitement in Arabic, Japanese, or Portuguese, not produce a flat, literal equivalent.

CAMB.AI's BOLI model powers this contextual translation layer, adapting phrasing to match the emotional register of the original while respecting cultural norms in each target language.

Step 5: Clone the Commentator's Voice

Voice cloning replicates the original commentator's vocal characteristics, including pitch, tone, cadence, and speaking style, from a reference audio sample. The dubbed output sounds like the same person speaking a different language.

Why does voice identity matter for sports? Because fans build loyalty to commentators. A cricket fan who loves a particular commentator's style wants to hear that same voice in Hindi, not a generic synthetic voice. Voice cloning preserves that relationship across languages.

CAMB.AI's MARS8 model achieves 0.87 WavLM speaker similarity and 0.71 CAM++ similarity, a 38% improvement over the nearest competitor on the MAMBA benchmark. For live broadcasting scenarios where latency matters most, MARS-Flash delivers ~100ms time-to-first-byte, fast enough for real-time sports commentary.

Step 6: Apply Emotion Transfer to Dubbed Audio

Flat, monotone dubbing kills sports commentary. When a commentator's voice cracks with excitement during a last-minute goal, the dubbed version needs to carry that same intensity.

Emotion transfer preserves the emotional quality of the original commentary in every dubbed version. The AI detects shifts in excitement, tension, disappointment, and celebration in the source audio, then applies those same emotional contours to the synthesized voice in each target language.

For high-stakes broadcasts like championship finals or playoff matches, MARS-Instruct provides director-level emotion controls with 1.2B parameters, giving production teams granular control over the emotional delivery of each dubbed output.

Step 7: Deliver Multilingual Streams to Your Platform

The final step is distributing your multilingual commentary streams to viewers. DubStream, CAMB.AI's live streaming dubbing product, ingests SRT, RTMP, or HLS feeds and outputs multilingual streams simultaneously.

Each language stream runs as a separate audio track that your platform can serve to viewers based on their language preference. Fans select their preferred language in the player interface, and the stream switches instantly.

For VOD highlights, post-match recaps, and social clips, DubStudio handles on-demand dubbing with the same voice cloning and emotion transfer pipeline. A 90-second highlight clip can be localized into 15 languages in minutes, ready for distribution across YouTube, Instagram, X, and TikTok.

Live Commentary vs. VOD Commentary: Key Differences

Every Fan Deserves Commentary in Their Language

Sports are universal. Commentary should be, too. If your broadcasts reach audiences in multiple countries, you are leaving fans behind every time you stream in a single language.

CAMB.AI already powers multilingual commentary for NASCAR, Ligue 1, FanCode, and the Australian Open. The same workflow scales to any sport, any league, and any platform.

Get started for free →