Full-Screen UI Mode

The full-screen UI is the default mode for the lcyt CLI. It provides a rich, multi-panel terminal interface for live captioning sessions.

Starting Full-Screen Mode

lcyt          # opens full-screen UI automatically when no other flags are given
lcyt -f       # explicit flag
lcyt --fullscreen

On the very first run (no stream key configured), the CLI launches a short setup wizard instead of the full-screen UI. After entering your stream key, the full-screen UI starts.

Panel Layout

The screen is divided into four panels:

Sending Captions

Type your caption text in the Input Field and press Enter to send it immediately. The caption is delivered to YouTube and logged in the Sent Captions panel.

Loading a File

Use the /load command to load a plain-text script, then step through it line by line:

/load /path/to/script.txt

Once a file is loaded, the Text Preview panel shows the current line. Use the arrow keys to navigate and Enter to send the current line.

Commands

Type a slash-command in the input field and press Enter:

Batch Mode

Batch mode lets you queue multiple captions and send them as a group:

1. Type /batch to enable batch mode.
2. Enter each caption and press Enter — lines are queued but not sent yet.
3. Type /send (or press the batch-send shortcut) to deliver the whole batch at once.

Keyboard Shortcuts

Options

All connection options can be set once and are saved to ~/.lcyt-config.json:

lcyt --stream-key YOUR_KEY   # Set stream key (saved for future sessions)
lcyt --base-url URL          # Override ingestion URL
lcyt --region reg1           # Set region identifier
lcyt --verbose               # Enable verbose logging

Interactive Mode

Interactive mode reads captions line by line from standard input. It is useful for scripting, piping text from another program, or captioning in a terminal without the full-screen UI.

Starting Interactive Mode

lcyt -i
lcyt --interactive

Once started, the CLI waits for you to type a line and press Enter. Each line is sent to YouTube as a separate caption.

How It Works

$ lcyt -i
> Hello, welcome to the stream!       ← you type this, then Enter
✓ Caption sent [seq 1]
> Today we will be discussing...
✓ Caption sent [seq 2]

• Each line you enter is sent immediately after pressing Enter.
• The sequence counter increments automatically.
• Press Ctrl+D (EOF) or Ctrl+C to quit.

Reading from a File or Pipe

Because interactive mode reads from stdin, you can pipe any program’s output directly into it:

# Send all lines of a text file, one by one
cat script.txt | lcyt -i

# Send output of a real-time transcription tool
my-transcriber | lcyt -i

When stdin is a pipe (not a TTY), the CLI reads lines until EOF and then exits automatically.

Options

Example: Scripted Session

#!/bin/bash
# Send a series of pre-written captions with a delay between each
captions=(
  "Welcome to the show!"
  "Today's topic: live captioning"
  "Stay tuned for more"
)

for caption in "${captions[@]}"; do
  echo "$caption" | lcyt -i
  sleep 5
done

Notes

• Interactive mode does not display the full-screen blessed UI; it is purely text-based and safe to use in CI pipelines, cron jobs, or scripts.
• The stream key must be configured either via --stream-key or stored in ~/.lcyt-config.json beforehand.
• Timestamps default to the current system time unless overridden with --timestamp.

Single Caption Mode

Single caption mode sends one caption (or a batch of captions) and exits immediately. It requires no interactive input and is the simplest way to send a caption from the command line or a script.

Sending a Caption

Pass the caption text as a positional argument:

lcyt "Hello, world!"
lcyt "Welcome to the stream!"

The caption is sent once and the process exits with code 0 on success or 1 on failure.

Setting the Stream Key

If you have not configured a stream key yet, set it with --stream-key. The key is saved to ~/.lcyt-config.json for future use:

lcyt --stream-key YOUR_KEY "First caption with key"

After the key is saved you can omit it in subsequent calls:

lcyt "Second caption"

Timestamp Override

By default the current time is used. Supply an explicit ISO timestamp with -t:

lcyt -t "2024-06-01T10:00:00.000" "Caption with a custom timestamp"

Batch Mode

Queue multiple captions across separate invocations and send them all at once:

# Queue captions (stored in ~/.lcyt-config.json)
lcyt -b "Line one"
lcyt -b "Line two" -t "2024-06-01T10:00:05.000"
lcyt -b "Line three"

# Send the queue in a single request
lcyt --send

The --send flag delivers all queued captions and clears the queue.

Heartbeat / Connection Test

Send a heartbeat to verify that the connection and stream key are working without sending a real caption:

lcyt --heartbeat

A successful heartbeat prints a confirmation and exits with code 0.

All Options

Exit Codes

LCYT — Live Captions for YouTube

LCYT (Live Captions for YouTube) lets you send real-time closed captions to a YouTube Live stream directly from your browser, using Google’s official HTTP POST caption ingestion API.

It is especially useful for non-English broadcasts — you can send captions in any language to make your stream accessible to a global audience.

What you can do with LCYT

Dashboard

The main dashboard is split into two panels side by side (or stacked on mobile):

• Left panel — Drop zone for script files, file tabs, the caption preview, and the microphone / audio meter
• Right panel — Log of all captions sent during the current session

The status bar at the top contains five buttons:

The input bar at the bottom is where you type and send captions.

Project components

Next steps

• Getting started — set up your stream and API key
• Sending captions — all the ways to send captions
• Caption settings — the CC modal: receivers, speech recognition, details, and translation
• General settings — the Settings modal: connection, theme, relay, and stream
• Status & Actions — the Controls panel
• Keyboard shortcuts — full shortcut reference
• Embedding in another site — iframe widgets for integration into production tools and custom dashboards

Getting Started

This guide walks you through the full setup from a brand-new account to your first live caption.

Prerequisites

Before you start you need:

1. A YouTube Live stream with a 30-second delay enabled
2. An LCYT API key (free — see below)
3. A modern browser (Chrome, Edge, or any Chromium-based browser recommended)

Step 1 — Enable closed captions in YouTube Studio

1. Go to YouTube Studio → Go Live.
2. In your stream settings, open the Advanced tab.
3. Enable Closed captions and choose the HTTP POST method.
4. Set the stream delay to at least 30 seconds (required by the caption API).
5. Copy your Stream Key — you will need it shortly.

Step 2 — Get a free API key

Visit lcyt.fi/app and fill in the free-key form with your name and email address. Your key is emailed to you instantly and works for up to 200 captions per day for 30 days.

Tip: Paid keys with higher limits and longer validity are available — contact the LCYT team.

Step 3 — Open the web app

Navigate to app.lcyt.fi.
On the first visit the Privacy modal opens automatically. Read the policy and click Accept to continue.

Step 4 — Enter your credentials

Click Settings in the top status bar to open the Settings modal.

Fill in the Basic tab:

Your credentials are saved automatically as you type. Close the modal when done.

Optionally enable Auto-connect to reconnect automatically on every visit.

Step 5 — Connect

Click the Connect button in the top status bar.

• The button turns green when the session is established.
• To disconnect, click the button again (it turns red on hover to confirm).

Step 6 — Send your first caption

Type a test caption in the input bar at the bottom of the page and press Enter (or click Send).

The caption appears in the Sent log on the right and is delivered to your YouTube stream within a few seconds.

Troubleshooting

Sending Captions

LCYT gives you several ways to get text onto the stream. You can use them in any combination during a session.

The input bar

The input bar sits at the bottom of the app. Type your caption text here and press Enter to send it immediately.

• Clear — remove the text without sending
• Send button — same as Enter

Tip: After sending, the input bar is cleared automatically so you can type the next caption right away.

Microphone (Speech-to-Text)

Click the microphone button (🎙) to start continuous speech recognition. Words are transcribed and displayed in the caption preview area in real time. The system sends each completed utterance as a caption automatically.

See Caption settings → Model tab for engine and language selection.

Desktop

The audio meter and mic toggle are embedded in the left panel.

Mobile

On phones, the mic button lives in the mobile audio bar at the bottom of the screen.

The mobile bar contains (left to right):

Loading a script file

You can load a pre-written caption script (plain text, one caption per line) by:

• Dragging and dropping the file onto the drop zone in the left panel, or
• Clicking the drop zone to open a file picker.

Once loaded, the file appears as a tab above the caption view. The current line is highlighted.

Navigating and sending file captions

Caption file format

Caption files are plain text (.txt), one caption per line. In addition to plain text, the file format supports metadata comments, stanza blocks, and empty-send markers.

Metadata comments

HTML-style comments on their own line attach metadata codes to all subsequent caption lines. Any key is accepted.

<!-- lang: fi-FI -->
<!-- section: chorus -->
<!-- speaker: Alice -->
<!-- lyrics: true -->
<!-- no-translate: true -->

To clear a code, set its value to empty:

<!-- lang: -->

Metadata comment lines are not sent as captions.

Stanza blocks

A stanza block attaches multi-line “singing aid” text to subsequent captions so viewers can see the upcoming lyrics. Open the block with <!-- stanza (no closing -->), write the stanza lines, then close with --> on its own line.

<!-- stanza
Amazing grace, how sweet the sound
That saved a wretch like me
-->
Amazing grace, how sweet the sound
That saved a wretch like me

The stanza text is pushed to the viewer when the first caption in the block is sent. To clear the stanza, add an empty block:

<!-- stanza
-->

Empty-send markers

A line containing only _ fires the current metadata codes (including the active stanza) to the viewer without sending any caption text to YouTube. This is useful for pushing a stanza to the viewer before the singing starts.

<!-- stanza
Amazing grace, how sweet the sound
-->
_
Amazing grace, how sweet the sound

You can optionally add a label after the underscore. The label is shown in red in the caption view as a visual cue for the operator — it is never sent to YouTube.

_ Show verse 1
_ ♪ Chorus
_ [pause here]

A bare _ displays a dimmed ⊘ send codes indicator. A labeled _ text displays the label text in red.

Batch mode

Batch mode lets you queue multiple captions and send them at regular intervals. This is useful for pre-written scripts where you want smooth, evenly timed delivery.

Configure the batch interval in Caption settings → Other tab.

Sent captions log

The right panel shows every caption you have sent in the current session, newest at the top. You can scroll back to review what was sent.

Caption Settings (CC)

Click CC in the top status bar to open the Closed Captions modal. It has up to four tabs, with the Details tab visible only in advanced mode.

Receivers tab {#receivers}

The Receivers tab is the first tab and manages all caption delivery destinations.

Click + Add target to add a new entry. Each entry can be:

You can add multiple targets (e.g. two YouTube streams and a viewer feed). Each target has an enable toggle. Only enabled targets receive captions.

Each target can also have Disable batch sending enabled, which forces captions to be sent individually to that target regardless of the global batch setting.

Add a YouTube target here with your stream key if you want to send captions directly to a YouTube Live stream.

Viewer targets

Viewer targets broadcast captions to anyone with the viewer page URL — no YouTube account required. After adding a viewer target you’ll see a shareable link:

https://app.lcyt.fi/view/<your-viewer-key>?server=https://api.lcyt.fi

You can also use the embeddable viewer widget:

https://app.lcyt.fi/embed/viewer?key=<your-viewer-key>&server=https://api.lcyt.fi

The viewer page displays the current caption prominently with a dimmed history list. It connects to the backend via Server-Sent Events and reconnects automatically on disconnect.

You can optionally assign an icon (PNG or SVG logo) to a viewer target. The icon is shown in the viewer page header. Upload icons in Settings → Icons (see below).

Service tab {#service}

The Service tab controls which speech-to-text engine is used and how it is configured.

STT engine

Microphone

Select which microphone to use if your device has more than one. Click Refresh to update the list.

Recognition language

Choose the spoken language for the speech recogniser. Type to filter the list.

STT model (Google Cloud only)

Additional Cloud STT options: Auto-punctuation, Profanity filter, Confidence threshold, Max caption length, and Google Service Account key upload.

Utterance end button (advanced mode only)

Shows a 🗣 icon on the audio meter during active speech recognition. Click it to force-end the current utterance immediately (commits the partial transcript as a final caption).

Utterance end timer (advanced mode only)

Automatically force-ends the utterance after N seconds (0 = disabled). Useful for segmenting long speeches into shorter captions.

Details tab (advanced mode only) {#details}

This tab is only visible when Show advanced options is enabled in Settings → Basic.

Batching

Transcription offset

Shifts the caption timestamp relative to when the transcription arrives. Use a negative value (e.g. −5 s) to compensate for transcription processing delay, so captions line up with the moment the speaker started talking in the YouTube stream.

Double-click the slider to reset to 0.

Client-side VAD

Voice Activity Detection (VAD) monitors microphone energy and forces the recogniser to finalise when silence is detected. Helps segment long unbroken speech on mobile Chrome.

Translation tab {#translation}

The Translation tab configures real-time caption translation.

Click + Add translation to add a target language. Each entry specifies:

Translation vendors

Show original

Enable Show original to include the original text alongside the translation in the YouTube caption stream (separated by a line break).

Translation

LCYT can translate your captions in real time and deliver the translation alongside the original text.

Click CC in the top status bar, then select the Translation tab.

Translation vendor

Adding a translation target

Click + Add translation to add a new row. Each row defines one translation output:

Target options

Note: Only one captions target is allowed at a time (YouTube accepts one caption block per request).

Multilingual video player

Every enabled translation language automatically becomes a selectable subtitle track in the embeddable video player at GET /video/:key — no extra configuration required. Viewers use their browser’s built-in CC button to switch languages.

See Multilingual Video Player for setup instructions and embed code.

Show original

Enable Show original to keep the original caption text in the stream alongside the translation. When disabled, only the translation is shown.

API key setup

Google Cloud

1. Create a project in the Google Cloud Console.
2. Enable the Cloud Translation API.
3. Create an API key and paste it into the Google Cloud API Key field.

LibreTranslate

1. Deploy a LibreTranslate instance (or use a public one).
2. Enter the server URL (e.g. https://translate.example.com).
3. Optionally enter an API key if your server requires one.

Settings

Click Settings in the top status bar to open the Settings modal.

Basic tab

The Basic tab contains all connection credentials and core preferences.

All fields are saved to your browser automatically as you type — no explicit Save button is needed.

Stream tab (advanced mode only)

This tab is only visible when Show advanced options is enabled on the Basic tab.

The Stream tab controls the RTMP relay and its associated broadcast settings.

Relay active toggle

Enable Active to start relaying your incoming RTMP stream to all configured destinations. When inactive, the backend accepts the stream but does not forward it. The toggle takes effect immediately — if your RTMP stream is already live when you enable the relay, fan-out starts without reconnecting.

Relay destinations

Click + Add relay to configure a new RTMP destination. Each entry supports:

Click ✕ to remove a destination. You can configure up to 4 destinations.

Per-slot advanced options (⚙ gear button)

Each relay slot has an optional advanced settings panel (click the ⚙ gear icon):

Transcoding (scale / FPS / bitrate) and CEA-708 caption mode cannot be combined on the same key — CEA-708 takes priority.

RTMP ingest address

When the relay is enabled on the server, the ingest address is shown at the bottom of the Stream tab:

rtmp://<server>/stream/<your-api-key>

Configure your broadcasting software (e.g. OBS) to push to this address. The stream key is your API key.

DSK RTMP ingest address

If the server supports DSK (Downstream Keyer) RTMP ingest, an additional address is shown:

rtmp://<server>/dsk/<your-api-key>

Push a second RTMP stream (e.g. a green-screen graphics feed) to this address to composite it on top of the main stream.

Requires an active backend connection and relay_allowed permission on your API key.

Controls Panel

Click Controls in the top status bar to open the Controls floating panel.

The Controls panel combines the session status display and diagnostic actions in one convenient location.

Status section

Stats button

Click Stats to open the Usage Stats modal. It shows per-key caption counts, daily limits, and session history. Requires an active connection.

My Files button

Click My Files to list caption and translation files saved on the backend for your API key, with download and delete options. Requires an active connection.

Actions section

Caption codes

Active metadata codes sent alongside every caption. Click a code button to toggle it:

File-level metadata (<!-- lang: fi-FI -->) overrides per-line codes.

File actions

Keyboard Shortcuts

LCYT is designed for fast, keyboard-driven operation during a live broadcast.

Global shortcuts (always active)

Shortcuts active when no text input is focused

These shortcuts are blocked when you are typing in any text field, to avoid accidentally sending captions or navigating while editing.

Sending

File navigation (when a script file is loaded)

Modals

Shortcuts reference in the app

All keyboard shortcuts are listed in the Settings → Credentials & Shortcuts tab for quick in-app reference.

Tips

• Keep the input bar unfocused (click anywhere outside it) to keep navigation shortcuts active.
• On mobile, use the navigation buttons in the mobile audio bar (−, ►, +) for the same line-navigation actions.
• Combine file navigation with speech recognition: navigate to a line and read it aloud while the mic is active, or manually advance through your script and send with Enter.
• Use Ctrl+1 to quickly jump to your first script file, Ctrl+2 for the second, and so on — handy when managing multiple scripts for different segments.

Embedding lcyt-web in Another Site

lcyt-web ships eight standalone embed widgets that you can drop into any page as <iframe> elements. Each widget renders only a specific part of the UI, configured entirely through URL parameters — no React knowledge required on the host site.

Overview

Looking for the video player? The embeddable multilingual video player is served directly from the backend, not from lcyt-web. Use <iframe src="https://api.example.com/video/<key>"> — see Multilingual Video Player.

Live iframe examples

Open any widget in a standalone iframe preview (opens a new window):

Audio capture example

Input + log example

Sent log example

File drop example

Files manager example

Settings example

RTMP relay example

Viewer example

URL Parameters

All embed pages share these common URL parameters:

When both server and apikey are present the widget connects automatically on load.

EmbedSentLogPage (/embed/sentlog) does not connect to the backend itself and therefore ignores server and apikey. It receives those from a sibling widget on the same page via BroadcastChannel.

Widget Details

/embed/audio — Audio Capture Widget

Renders the full AudioPanel (microphone button, interim text, VAD, translation pipeline). Owns the backend session when server + apikey are supplied.

<iframe
  src="https://your-lcyt-host/embed/audio?server=https://api.example.com&apikey=YOUR_KEY&theme=dark"
  allow="microphone"
  style="width:100%; height:220px; border:none;">
</iframe>

The allow="microphone" attribute is required for speech recognition to work inside an iframe.

/embed/input — Input Bar + Sent Log Widget

Renders a text caption input field at the bottom and a scrollable sent-captions log above it. Owns its own backend session.

<iframe
  src="https://your-lcyt-host/embed/input?server=https://api.example.com&apikey=YOUR_KEY&theme=dark"
  style="width:100%; height:320px; border:none;">
</iframe>

/embed/sentlog — Sent Log Widget (read-only)

Renders only the sent-captions log. Does not own a session. Instead, it receives the session JWT token and caption texts from a sibling /embed/audio or /embed/input widget on the same host page via the browser’s BroadcastChannel API, then opens its own independent EventSource connection to /events on the backend to receive real-time delivery confirmations.

<iframe
  src="https://your-lcyt-host/embed/sentlog?theme=dark"
  style="width:100%; height:320px; border:none;">
</iframe>

Requirement: At least one other embed widget (/embed/audio, /embed/input, /embed/file-drop, or /embed/files) must be present on the same host page and connected to the backend.

/embed/file-drop — Simple File Drop Widget

The minimal caption-from-file widget. Phase 1 shows a large drag-and-drop zone (click to browse is also supported). Drop one .txt file and Phase 2 immediately shows the player:

• The current line is displayed prominently in the centre of the widget.
• ◀ Prev, Send, ▶ Next buttons control navigation and delivery.
• Keyboard shortcuts work without clicking buttons: ↑ / ↓ to move, Enter to send and advance.
• A ✕ reset link in the header returns to Phase 1 to load a different file.
• The connection status dot in the header shows whether the backend relay is connected.

Send delivers the current line to YouTube and automatically advances the pointer to the next line.

<iframe
  src="https://your-lcyt-host/embed/file-drop?server=https://api.example.com&apikey=YOUR_KEY&theme=dark"
  style="width:100%; height:280px; border:none;">
</iframe>

/embed/files — Full File Management Widget

Renders the complete file-based captioning workflow from the main app, without the status bar, settings modals, or audio panel:

• FileTabs row at the top — switch between open files, add new files, toggle drop zone.
• DropZone — collapsible drag-and-drop file loader (auto-hides once a file is loaded).
• CaptionView — scrollable line list with the active pointer highlighted; supports raw text editing.
• InputBar — text input with batch mode, translation, and all keyboard shortcuts.
• Sent log panel — togglable delivery log (✓✓ button in the toolbar); starts visible by default.

Line double-click in CaptionView sends that line immediately via the InputBar, exactly as in the main app.

<iframe
  src="https://your-lcyt-host/embed/files?server=https://api.example.com&apikey=YOUR_KEY&theme=dark"
  style="width:100%; height:640px; border:none;">
</iframe>

To start with the sent log panel hidden, add &sentlog=0 to the URL.

/embed/settings — Settings Widget

Renders the Settings panel as a standalone widget. Provides the General tab (backend URL, API key, stream key, theme) and the CC tab (caption targets, STT language, translation settings). The widget connects automatically if credentials are present in the URL.

<iframe
  src="https://your-lcyt-host/embed/settings?server=https://api.example.com&apikey=YOUR_KEY&theme=dark"
  style="width:100%; height:480px; border:none;">
</iframe>

Useful for building custom operator dashboards where settings management and captioning are in separate panels.

/embed/rtmp — RTMP Relay Widget

Renders the RTMP relay slot management UI as a standalone widget. Shows the relay active toggle, configured slots, RTMP ingest address, and per-slot advanced options (scale, FPS, bitrate, caption mode).

<iframe
  src="https://your-lcyt-host/embed/rtmp?server=https://api.example.com&apikey=YOUR_KEY&theme=dark"
  style="width:100%; height:320px; border:none;">
</iframe>

Requires an active backend connection and relay_allowed on the API key.

/embed/viewer — Caption Viewer Widget

Renders a read-only live caption display for audience members. Connects directly to the backend’s public GET /viewer/:key SSE endpoint — no API key or JWT required.

<iframe
  src="https://your-lcyt-host/embed/viewer?key=my-event-key&server=https://api.example.com&theme=dark"
  style="width:100%; height:200px; border:none;">
</iframe>

URL parameters specific to /embed/viewer:

This widget does not need apikey or a session — it is intended for the audience, not the operator. The server and key parameters are required; the widget shows a “waiting for stream” state until captions arrive.

The full-screen version (non-embed) is available at /view/:key?server=<backendUrl>.

Cross-Widget Communication

When widgets are spread across different parts of the same host page they coordinate using the browser’s BroadcastChannel API (channel name: 'lcyt-embed'). All iframes must be served from the same origin (the same lcyt-web deployment URL) for BroadcastChannel to work — this is a browser security requirement.

Message types

The sentlog widget also opens its own EventSource to GET /events?token=... on the backend to receive caption_result and caption_error events independently, without going through the session-owning widget.

Splitting the UI Across a Host Page

A common integration pattern is to place each widget in a different region of the host page:

<!-- Left sidebar: microphone -->
<div id="sidebar-left">
  <iframe
    src="https://your-lcyt-host/embed/audio?server=https://api.example.com&apikey=YOUR_KEY"
    allow="microphone"
    style="width:100%; height:240px; border:none;">
  </iframe>
</div>

<!-- Main content: text input -->
<div id="main-input">
  <iframe
    src="https://your-lcyt-host/embed/input?server=https://api.example.com&apikey=YOUR_KEY"
    style="width:100%; height:300px; border:none;">
  </iframe>
</div>

<!-- Right sidebar: delivery log (read-only, no credentials needed) -->
<div id="sidebar-right">
  <iframe
    src="https://your-lcyt-host/embed/sentlog"
    style="width:100%; height:400px; border:none;">
  </iframe>
</div>

Only one widget should own the session at a time. If both /embed/audio and /embed/input are on the page with the same credentials they will each start a separate session — captions from one will not appear in the other’s log unless you use /embed/sentlog (which listens to all sibling broadcasts). For a split audio + log layout, use /embed/audio as the session owner and /embed/sentlog for the delivery log.

CORS Configuration

The backend must accept requests from the origin of the host page (not the lcyt-web origin). Configure the ALLOWED_DOMAINS environment variable on the backend to include the host site’s domain:

ALLOWED_DOMAINS=lcyt.fi,www.lcyt.fi,yoursite.com,www.yoursite.com

The embed widgets themselves are served from the lcyt-web origin and communicate with the backend using the API key — the CORS domain is the domain registered when the API key was created, which must match the domain field sent in POST /live.

Listening for postMessage Events (Optional)

All embed pages that own a session (/embed/audio, /embed/input) are ready for future postMessage integration with the host page. Currently, the BroadcastChannel mechanism handles cross-widget communication automatically. If you need the host page itself to react to caption events you can listen on the BroadcastChannel from a host-page script:

const ch = new BroadcastChannel('lcyt-embed');
ch.onmessage = (ev) => {
  if (ev.data.type === 'lcyt:caption') {
    console.log('Caption sent:', ev.data.text);
  }
  if (ev.data.type === 'lcyt:session') {
    console.log('Session token received');
  }
};

Note: BroadcastChannel only works between same-origin contexts. The host page script must be served from the same origin as the embed iframes (i.e., the lcyt-web deployment URL), or you must use the iframes’ own postMessage API relayed through the parent page.

Multilingual Video Player

When you configure a viewer target in the CC → Targets tab, the backend automatically generates an embeddable video player at:

https://api.example.com/video/<viewer-key>

This player combines your live video stream with real-time subtitle tracks in every language you have configured under CC → Translation. Viewers can select their preferred language using their browser’s built-in CC (closed captions) button — no custom controls are needed.

What you need

1. HLS video stream — the RTMP stream you send to the server must have hls_enabled = true on your API key (set by an admin). The video and subtitle tracks share the same key.
2. Viewer target — configure a viewer target in CC → Targets with the same key you use for the HLS stream. Every caption you send is automatically transcribed into subtitle segments.
3. Translations (optional) — add one or more translation languages in CC → Translation. Each language appears as a separate subtitle track in the player.

Setting up

1. Configure a viewer target

Open CC → Targets tab and add a target:

Use the same key for the HLS video stream (the stream key configured in nginx-rtmp must match).

2. Add translation languages (optional)

Open CC → Translation tab and add any languages you want as subtitle tracks. Each enabled language will appear in the player’s CC menu as a selectable track.

The original caption text is always included as the Original track regardless of whether any translations are configured.

3. Share the player URL

The player is available at:

https://api.example.com/video/<viewer-key>

Replace api.example.com with your backend URL and <viewer-key> with the key you set in step 1.

Embedding the player

The player page is iframe-embeddable on any website:

<iframe
  src="https://api.example.com/video/my-event-2026"
  width="960" height="540"
  frameborder="0"
  allow="autoplay"
  allowfullscreen>
</iframe>

Theme

Add ?theme=light for a light background:

<iframe src="https://api.example.com/video/my-event-2026?theme=light" ...></iframe>

Subtitle language selection

The player exposes all active subtitle tracks through the browser’s native CC button in the standard video controls:

• Chrome / Edge / Firefox — click the CC (⧉) button in the video controls to open the language menu.
• Safari (macOS / iOS) — click the CC button or the subtitles option in AirPlay / fullscreen controls.
• Android Chrome — tap the CC button in the controls.

Tracks appear as soon as the first captions arrive for that language — you do not need to reload the player.

How subtitles are generated

Each time a caption is sent, the backend:

1. Writes the original text and each translation into rolling 6-second WebVTT segment files (one set per language).
2. Updates an HLS subtitle playlist per language, using EXT-X-PROGRAM-DATE-TIME headers so the player can align subtitle cues to the video by wall clock.
3. Serves a master HLS manifest that references both the video stream and all subtitle playlists.

Because the subtitle system uses the same viewer key as the SSE viewer endpoint (/viewer/:key), there is no extra configuration — any key that receives captions automatically gets a subtitle sidecar.

Subtitle track reference

Troubleshooting

Technical details

For the full API reference, including manifest format, WebVTT segment format, and environment variables, see API: /video.