How to Run a Multilingual Live Event with Remote Interpreters

Simultaneous interpretation used to mean booths, a hardware distribution system and a technician for every channel. For an online audience, most of that disappears. With DCAST, each interpreter can work from home with a laptop or a phone and a headset, sending their voice to their own language track; viewers watch one stream and choose their language in the player.

The technology is the easy part. What makes an interpreted event work is planning: who hears what, when everyone rehearses, and what happens when something goes wrong. This guide is that plan.

The shape of the setup

Every language is a separate feed with its own key. Interpreters do not need to be in the same place, and they do not send video: DCAST uses only the audio of a language feed.

Roles you need

• Producer — runs the main encoder and the DCAST stream settings.
• Interpreting coordinator — the interpreters' single point of contact; monitors every language during the event and decides what to do if one fails.
• Interpreters — for events longer than about 30 minutes, professional practice is two interpreters per language taking turns. Decide in advance who hands over the key — only one of them should be connected at a time on a given track, sending from one encoder.
• Moderator — in the chat, answering "I can't hear Spanish" in a friendly way.

What each interpreter needs

• A computer or phone with a stable connection. A language feed is audio plus, at most, a tiny video — well under 1 Mbps upload.
• A headset with a microphone, so they hear the speaker and do not send the speaker back through their microphone.
• An encoder: Larix Broadcaster on a phone (audio-only mode) or OBS Studio on a computer with a tiny canvas. Our step-by-step guide covers both, plus vMix.
• Their track's connection details from DCAST — and only theirs.
• A return feed: a clean, low-delay source of the original sound (see next section).
• A backchannel to the coordinator — a messaging group is enough.

Recommended settings for every language feed: AAC, 48 kHz, 128–192 kbps, video none or as small as the encoder allows.

The return feed: how interpreters hear the speaker

This is the part most first-time organisers get wrong. Do not ask interpreters to listen to the public stream. A public live stream is intentionally delayed — often by many seconds, so an interpreter listening to it will always be behind — and their interpretation will reach viewers even later.

Better options, from best to simplest:

1. A low-delay audio link from the production — for example a dedicated call in a conferencing tool where the production shares the program sound and interpreters join muted.
2. The interpreter is in the room or on the venue's audio system.

Whatever you choose, test that interpreters hear the speaker clearly and without their own voice coming back to them.

Setting up the languages in DCAST

1. Open the stream's settings → Feeds → Languages.
2. Add language once for each language; rename each track to what viewers should read, such as "Français" or "Українська".
3. Send each interpreter their track's details. For an RTMP stream, that is the server URL shown in the Languages tab plus the track's stream key; for an SRT stream, the track's SRT URL or Stream ID.
4. Up to 16 languages per stream are available. Create all of them before you press Start.

Language feeds can connect before, together with or after the main stream, so interpreters can be connected and waiting before you go live. The Languages tab shows a signal dot per track: green while that language's feed is arriving.

Rehearsal timeline

One week before

• Send every interpreter their key, the recommended settings and the encoder guide.
• Hold a 15-minute technical check with each interpreter: they connect, the dot turns green, you hear them in the player.

The day before

• Full run-through with the real production: main encoder, return feed, all interpreters.
• Open the watch page on a phone and a computer, switch through every language, check levels against Original.
• Agree the backchannel rules: who speaks to whom, and how a failing interpreter hands over.

One hour before

• Everyone connected, every dot green.
• The coordinator listens to each language for 30 seconds.
• Post a short note on the event page: "Choose your language with the language button in the player" — or link our viewer guide.

Event-day rules

• Run the event as one continuous broadcast. The replay keeps the languages, but if you press Stop and Start again on the same stream, it offers the extra languages only for the part after the last Start. For breaks, keep the stream running with a holding picture and background music.
• Interpreters stay connected through breaks, muted if they wish, rather than disconnecting.
• If an interpreter drops, they reconnect with the same details; the coordinator tells the chat moderator, who reassures viewers of that language.
• The coordinator listens to everything. Viewers rarely report a silent language quickly; the coordinator should notice first.

After the event

The replay of the stream keeps the languages: people who missed the live event choose their language with the same button in the DCAST player. If you also want separate files — for example to publish an interpreted version elsewhere — ask each interpreter to record their own audio locally (OBS and most encoders can record while streaming), and keep a local recording of the program on the main encoder.

Checklist

• One track per language created and named in Feeds → Languages
• Each interpreter has only their own key or Stream ID
• Every language feed: AAC, 48 kHz, 128–192 kbps, minimal or no video
• Interpreters wear headsets and have a low-delay return feed
• Backchannel to the coordinator works
• Rehearsal: every dot green, every language heard in the player
• Plan for breaks without pressing Stop
• Local recordings running on the main encoder and on each interpreter's side