Frequently asked questions about the AI interpreter

It is an automated simultaneous interpretation system that captures the speaker's audio, processes it in real time with language models specialized in interpretation, and delivers the translation as audio or captions. It was trained on real lingo booth transcripts across 24 sectors, which improves terminology in areas such as finance, energy, healthcare and technology.

The speaker's audio enters the system, is transcribed, interpreted and synthesized into the target language, all in a continuous flow. The AI works with 30 to 60 seconds of context windows to decide how to build each sentence, much like a human interpreter who hears the start of an idea before translating it.

For remote events (Zoom, Teams, Meet) typical latency is 2 to 4 seconds, and can reach 3 to 5 seconds on Google Meet. For in-person events, it is between 2 and 6 seconds. That is comparable to a human interpreter in simultaneous mode, and most participants adapt within a few minutes.

Not in every scenario. For structured presentations, training and internal events, AI delivers good results at a lower cost. For negotiations, M&A, boards, technical due diligence and arbitration, lingo recommends a human interpreter. The choice depends on how critical the content is, not only on the language.

Automatic captions show text on a screen and the participant has to read. The AI interpreter delivers the translation as audio, in the chosen language channel, as a human interpreter would. Both can run together at the same event, and lingo also offers live captioning as a separate service.

The highest-quality pairs are EN↔PT, ES↔PT, FR↔PT and DE↔PT. For Asian languages such as Japanese, Chinese and Korean, quality is lower and we recommend a human interpreter. Before confirming a less common pair, lingo evaluates the type of content and recommends the best option.

No. In presentations with a deck and script, especially EN↔PT, the difference is small. In free Q&A, very fast speech, strong regional accents and highly specific terminology, the human is clearly better. That is why lingo only recommends AI where that difference is acceptable for the event.

It understands common accents, but accuracy drops with very strong accents, overlapping speech and a fast pace. To improve results we ask speakers to keep a steady pace, use a microphone close to the mouth and avoid interruptions. In debates with many simultaneous speakers, a human is still the best choice.

Yes. Before the event, lingo loads the client's glossary into the system: internal acronyms, product names, proprietary terms and executive names. The more material you send in advance (slides, scripts, reports), the higher the accuracy on your company-specific vocabulary.

Numbers, monetary values and acronyms are sensitive points in any interpretation. With the glossary loaded and the slides in hand, the AI gets most cases right, but for critical data such as transaction values or financial metrics we recommend human review or a dedicated human interpreter.

This is a known limit. The AI tends to translate more literally, and humor, irony and idiomatic expressions can lose their effect. In motivational talks, relationship events and speeches with strong cultural content, a human interpreter adapts the tone and preserves the intent.

Only to a limited extent. The synthetic voice is clear and steady but does not reproduce all of the speaker's emotional nuance. For training, announcements and informational content this is rarely a problem. In leadership speeches or presentations where emotion is part of the message, a human delivers more.

A lingo technician monitors the AI channel for the full duration of the event and can intervene immediately if quality falls below the acceptable threshold. In the hybrid setup, a human interpreter stands by and takes over the channel without a noticeable interruption for participants.

Yes. The AI joins as a co-host with Language Interpretation permission, occupying the same slot a human interpreter would. The organizer enables the channel under Settings → Language Interpretation, and each participant picks a language in their own app. Setup takes about 15 minutes before the meeting.

Yes. The Language Interpretation feature requires Zoom Pro, Business or Enterprise. The free plan does not support it. If you are not sure about your plan, lingo checks with you before the event and tells you what needs to be adjusted.

On Teams, the integration runs through an interpretation bot in the meeting channel and requires a Teams Premium license, with about 20 minutes of setup. On Google Meet Enterprise, the integration uses an interpretation API, with slightly higher latency of 3 to 5 seconds.

No. On supported platforms, participants simply select the language in the interpretation selector of Zoom, Teams or Meet and hear the AI exactly as they would hear a human interpreter. There is no extra app and no sign-up.

Yes, and it is a very common setup. The AI acts as the primary interpreter and a human interpreter stands by in the same channel. If the AI fails or quality drops, the human takes over without participants having to switch channels.

Yes, and it is one of the best uses. The AI does not get tired, which helps in events of 4 hours or more and in meetings at hard hours, where scheduling human interpreters is more complicated. We keep technical monitoring for the whole duration.

lingo connects the system to the venue's sound feed through a professional audio interface. The translation goes out through participants' individual UHF receivers or a caption screen on stage. Processing can be local, on dedicated hardware, without sending audio to the cloud.

In-person setup takes about 90 minutes before the event starts, including connection to the audio system, channel tests and loading the glossary. At venues with complex audio, lingo talks to the venue's AV team in advance.

If processing runs in the cloud, yes: we recommend at least 50 Mbps dedicated. If the event requires local processing, lingo uses a preloaded model on dedicated hardware, which removes the internet dependency for the interpretation itself.

The output can go to the existing UHF transmitter or to caption screens over HDMI. If the venue has no equipment, lingo supplies receivers, transmitter and a technical operator in the same quote.

We do not recommend it for M&A negotiations, private equity and venture capital boards, technical due diligence, IPO roadshows or arbitration. In those contexts an imprecision can have financial or legal consequences, and a human interpreter is the standard.

In-person AI costs between 40% and 60% of a human interpreter for half-day or full-day events. Remote AI typically costs 30% to 50% less for events of two hours or more. For very short events the difference is small. Request a quote for figures that fit your case.

By default, no. Processing happens in real time and audio is not recorded. For clients who need an explicit no-storage guarantee, lingo offers 100% local, offline processing, and this can be written into the contract.

It depends on the level of secrecy and your legal team's requirements. With local processing and no recording, exposure risk drops considerably. For highly sensitive topics such as M&A transactions, lingo recommends a human interpreter under a confidentiality agreement.

Request a quote in 60 seconds through the website assistant or message Alex on WhatsApp. Provide the date, duration, languages, format (in person or remote) and platform. lingo will tell you whether AI, human or a hybrid setup suits your event best.

Yes. For recurring or larger events, lingo runs a short test with samples of your content so you can assess quality and adjust the glossary before the event day.

An interpretation company. lingo has operated since 2005, with 21 years of experience in corporate events, and added AI as another service option, always with technical supervision and a human backup interpreter when the event requires it.

Product launches with a defined script, standardized corporate training, onboarding, compliance, operational meetings between branches, webinars and long events where fatigue would be a problem. In all of them the content is more predictable and the cost of an occasional imprecision is low.

It works well for sessions with structured presentations and slides. For panels with open debate and audience questions, we suggest combining: AI for keynote talks and a human interpreter for panels and Q&A. lingo designs the event session by session to balance cost and quality.

Yes. It is common to use a human interpreter for critical sessions, AI for the rest and live captioning for accessibility and event records. lingo configures the channels in parallel and one technician coordinates the entire operation from start to finish.

Yes. You can hire the AI interpreter for specific sessions, such as training blocks or product demos, and keep human interpreters for sensitive discussions. lingo plans the schedule so the handover between channels is smooth for the audience.