Live Subtitles

Turn live speech into multilingual subtitles as it happens.

Live Subtitles recognizes speech from the audio track, creates subtitles in the language being spoken, and translates them into selected languages. The resulting subtitle tracks or sidecar subtitle outputs are delivered alongside the live stream.

Discuss your subtitle workflow Explore common workflows
Live speech recognition
Follows the language heard in the live audio
Flexible output
Subtitle tracks or sidecar subtitle outputs
CloudTV PlatformLive subtitle route
Routes · Languages · Subtitle outputs · Health
Live streamVideo + audio
Video track
  1. Audio trackLive speech
  2. AISpeech recognition
  3. Spoken languageLive subtitles
  4. OptionalLanguage translation
Route outputSubtitle tracks or sidecar outputsDelivered with the Live Media Core output
Live Media Core activeSpeaker context maintainedRoute monitored

Real-world workflows

Help more viewers follow the same live stream.

Add subtitles for the language being spoken, translate them for an international audience, or connect subtitle output to the live stack you already operate.

01 · International audiences

Reach viewers in more languages from one live stream.

When one speaker needs to reach audiences in several languages.

A creator, lecturer, presenter, or event guest may speak one language while viewers join from many countries. Running a separate live stream or arranging a separate human subtitle workflow for every language is expensive and difficult to coordinate.

The speech is recognized once in the language being spoken. Live Subtitles keeps that subtitle track and adds the configured target languages from the same recognition result, allowing viewers to choose from the available languages in a compatible player.

Best suited to International creator broadcasts, global online events and product launches, international media, and multinational company streams.
  1. One live stream
  2. Speech recognition
  3. Spoken-language subtitles
  4. Selected translations
  5. Viewer selects a language
02 · Spoken-language subtitles

Make live speech easier to follow as it happens.

When viewers need subtitles without translation.

Speech rate, accents, environmental noise, or listening conditions can make meetings, classes, lectures, and interviews harder to follow. The workflow may only need timely subtitles for what is being said, not a translated version of the stream.

With translation disabled, Live Subtitles follows the spoken audio and produces subtitles in the language being spoken. The result is delivered as the configured subtitle track or sidecar output to the existing live player.

Best suited to Online education, public lectures, video meetings, webinars, public speaking, and live interviews.
Live spoken audio
Speech recognition
Subtitles matching the spoken languageTrack or sidecar output

Automatic live subtitles can help viewers follow spoken content, but this page does not claim certification under a specific accessibility regulation.

03 · Keep existing delivery

Add subtitles without rebuilding the delivery workflow.

When the existing live stack has no automatic subtitle path.

You may already operate the encoder, CDN, publishing platform, and player and have no reason to migrate the whole delivery system simply to add live subtitles.

Route the live source through Live Media Core with Live Subtitles, then send the resulting live output back to your existing CDN, platform, player, or application. Global Edge CDN is not required.

Best suited to Established live platforms, broadcasters, online education and meeting providers, and services that use a third-party CDN.
  1. Existing live source
  2. Live Media Core+ Live Subtitles
  3. Existing CDN or platform
  4. Existing player or application
04 · Output form

Deliver subtitles as tracks or sidecar outputs.

When the destination expects a particular subtitle form.

Some players read subtitle tracks carried with the live output, while other platforms retrieve subtitles from a separate sidecar address.

Configure the live route to deliver its complete subtitle language set as subtitle tracks or sidecar outputs, then connect destinations that accept that form. The route does not assign a different language set or output form to each destination, and the live speech is not recognized again for every downstream system.

Best suited to Multi-platform publishing, services with both native and web players, CDN and platform integrations, and live services that let viewers choose a subtitle language.
One recognized subtitle setSpoken language + selected translations
Subtitle tracksCarried with the live output
or
Sidecar outputsRead from separate addresses

Destinations connect to the output form configured for the route.

Languages and output

Turn live speech into the subtitle set the route needs.

Live Subtitles follows the speech in the live audio without requiring a source-language setting. When translation is disabled, the subtitles remain in the language being spoken. When translation is enabled, subtitles are also produced in the configured target languages.

CloudTV PlatformTranslation and language setNo source-language setting is required.
Live streamSpeech in the live audio
AI speech recognitionOne shared recognition result
Spoken-language subtitles onlySubtitles follow the language heard in the live audio.No source-language selection or translation stage is required.

Keep subtitles in the language being spoken. No source-language setting is required. Speech recognition follows the language heard in the live audio and produces the corresponding subtitles without presenting language detection as a separate customer configuration step.

One output form per live route

Choose how the complete language set leaves the route.

A compatible player can display and switch between configured subtitle tracks, while a sidecar workflow reads the same language set from its subtitle output. Different destinations cannot currently select a different subtitle form on the same route.

Subtitles are not burned into the video by default. Sidecar subtitles do not require the video to be re-encoded, and repackaging a subtitle track for a compatible output protocol is not a video transcode.

Route language setSpoken · EN · ES · JA
Live output with subtitle tracksOne compatible live output carries the configured subtitle set.
DestinationsCompatible players and platforms

Live Subtitles processes the active live route. CloudTV does not retain a subtitle archive or a complete transcript as part of this product.

Speaker and context

Follow the speaker. Preserve the meaning.

Live Subtitles does more than convert isolated phrases into text. As the conversation continues, the system follows speaker turns and maintains context across the live session. Translations use that context to preserve meaning, references, and terminology instead of treating every subtitle cue as a separate sentence.

Live speech
AI speech recognition
Speaker turnsConversation contextShared recognition result
Spoken languageSubtitles matching the speech
Selected languagesContext-aware translation
Speaker turns

Follow a conversation as participants change.

Distinguish who is speaking as a live conversation moves between participants. This describes continuity between speakers, not identity verification or voiceprint authentication.

Conversation context

Keep references and terminology consistent.

Use the conversation already heard to keep pronouns, references, proper names, and domain terminology consistent as the live session continues.

Translation context

Translate the continuing conversation, not isolated cues.

Translate from the shared recognition result and ongoing context rather than treating each subtitle cue as an unrelated sentence.

Quality is measured in separate dimensions

One score cannot describe the whole subtitle workflow.

Speech recognition
Compared with human-corrected text using WER or CER for each language.
Speaker continuity
Checked across conversations where several people take turns speaking.
Context continuity
Reviewed for references, pronouns, proper names, and terminology across the continuing conversation.
Translation quality
Reviewed by people who know the target language for meaning, context, terminology, and natural expression.
Subtitle timing
Measured for appearance time, duration, and alignment with live speech.
Live environments
Tested separately in studios, meetings, classes, interviews, and noisy environments.

Performance varies by language, accent, audio quality, noise, and specialist vocabulary. For legal, medical, or other workflows that require word-for-word certainty, automatic live subtitles should not be presented as a replacement for human-reviewed text.

Live latency

Keep subtitles close to the live moment.

Live Subtitles processes speech continuously while the stream is live. Compared with the same live route running without subtitles, enabling speech recognition and multilingual subtitle output adds less than two seconds of latency in the tested configurations.

Less than 2 secondsof added live-stream latency
Same sourceLive camera + microphone
Same ingestLive Media Core
BaselineSubtitles disabled
ComparisonLive Subtitles enabled
Same destinationPlayback endpoint
Operational latencyMeasured from the same live source to the same playback endpoint.
Less than two seconds of added latency in tested configurations
What changesSpeech recognition and the configured multilingual subtitle output are enabled
What stays the sameThe live source, media route, destination, and playback endpoint
What still affects total latencyThe encoder, output protocol, delivery network, and player

What the figure means. Less than two seconds is the additional processing delay observed when Live Subtitles is enabled on the same live route. It is not the route’s total end-to-end latency.

Fit your existing stack

Add Live Subtitles without replacing the rest of your live stack.

Live Subtitles runs with Live Media Core, but the rest of the route remains yours to choose. Keep your current CDN, publishing platform, and player, or add CloudTV delivery and applications where they fit.

SourceYour live sourceKeep the encoder and source workflow already in place
Required processingLive Media Core + Live SubtitlesSpeech recognition · Configured languages · Track or sidecar output
Existing deliveryYour CDN, publishing platform, or playerReturn the subtitle-enabled output to the systems you operate

Global Edge CDN is not required. Send the live output back into the systems you already operate.

CloudTV PlatformConfigureMonitorLanguagesOutput formRoute health
Required platform

CloudTV Platform

Target languages · Output form · Route health

Required processing path

Live Media Core

Live Subtitles runs inside the active live media route

Optional delivery

Global Edge CDN

Keep an existing CDN or enable CloudTV delivery

Optional applications

Native SDK & Applications

Subtitle display and language switching, or keep an existing player

Optional transport

Resilient Transport

Improve difficult ingest or playback network paths

Optional operations

Business Management

Content · Subscriptions · Customers · Devices

Optional protection

Rights Protection

Authorization · DRM · Invisible watermarking

An enabled Live Subtitles route always runs through Live Media Core. It is not a plug-in that runs directly inside a third-party media server. Global Edge CDN, Native SDK & Applications, Business Management, Resilient Transport, and Rights Protection remain optional.

Plan the subtitle output

Tell us where your subtitles need to go.

Share your live source, output protocol, target languages, and whether your destinations accept subtitle tracks or sidecar subtitles. We’ll help map the right live workflow through Live Media Core.

Discuss your subtitle workflow