KADENZ translates talks simultaneously into up to four languages, about six seconds behind the speaker. It needs no booth and works in any room. Your guests listen on our receivers, and our technician looks after the event. We developed KADENZ ourselves. In Bern, KADENZ is only available at Eventforum.
Up to four languages at once
About six seconds behind the speaker
Glossary with your names and acronyms
Receivers on site, no app needed
In any room, no booth needed
A fraction of the cost of interpreters
The speaker’s microphone feeds KADENZ, in the hall through our mixing desk. KADENZ recognises the speech, interprets it simultaneously and speaks each language on its own radio channel. Your guests select the channel on their receiver.
KADENZ takes the original audio straight from the mixing desk or the microphone. A language model conveys the meaning of each sentence, as interpreters do, and condenses when someone speaks fast. Each language has its own channel with a voice from ElevenLabs. The translation follows the speech about six seconds behind. Anyone who understands the speaker’s language simply listens to the room.
Original audio and three languages live
KADENZ does not need an interpreting booth. In each room we set up a laptop and one radio transmitter per language. KADENZ takes the sound straight from the microphone or the mixing desk. It therefore interprets in the event hall as well as in a workshop room, and in several rooms at the same time, for example in three workshop rooms with two languages each. If you bring your own team of interpreters, they work in our booths with Televic equipment.
In workshop rooms too
At the entrance, your guests receive a receiver and choose the channel for their language. 40 receivers are available.
Receivers are handed out at the entrance
A public talk in standard German, interpreted live. You first hear the speaker. A few seconds later the interpretation starts and the original moves into the background, just as your guests hear it in the hall.
Original audio: Knud Jahnke, “Euclid, das Weltraumteleskop – 180 Millionen Galaxien sind ein guter Anfang”, 38th Chaos Communication Congress, Hamburg 2024 (media.ccc.de), licence CC BY 4.0. Excerpt 1:06 to 2:09, mixed lower and overlaid with the KADENZ interpretation of 7 October 2026; the interpretation is machine-made and unedited. The speaker has no connection with Eventforum.
Planning a multilingual conference, general meeting or congress? We will show you how KADENZ sounds at your event.
Make an enquiryThere are three ways to run a multilingual event today. This is how they differ in the hall.
| Criterion | KADENZ at Eventforum | Translation app via QR code | Your own interpreters |
|---|---|---|---|
| How guests listen | On the receivers of the system, one channel per language, on request also on a phone | On their own phone with headphones | On the receivers of the system |
| Support | Our technician throughout the event | Mostly self-service, remote help | Your team, plus our technician for the system |
| Delay | About six seconds | Varies by provider | A few seconds |
| Voices | A female and a male voice per language, from ElevenLabs | Usually one or two voices | Human voices |
| Names and technical terms | Glossary set up with you in advance | Depends on the provider | Prepared by the team |
| Contract | Eventforum Bern, in Swiss francs | Usually a provider abroad | With the team, the system from us |
| Cost | A fraction of an interpreting team, plus the system | Licence per event or subscription | Fee per person and day, plus the system |
| Best for | Talks and presentations in standard German | Many languages, guests outside the hall | Debates, Swiss German, confidential content |
As of October 2026, based on public information from widely used providers.
Eventforum rents out the booths, the system and the receivers. KADENZ now interprets on request; your own team of interpreters works in our booths. We do not provide interpreters.
AI interpreting costs only a fraction of an interpreting team. In both cases you rent the system with booths and receivers as well.
For talks, presentations, specialist conferences and general meetings in standard German or English, in any room and without a booth. Up to four languages at once, a glossary with your names, about six seconds behind.
Better for Swiss German on stage, debates with interruptions, negotiations and confidential content. Your team works in our booths with Televic equipment; guests listen on the same receivers.
From enquiry to after the event in three steps. Our team is with you along the way.
We wanted AI interpreting that works in the hall like an interpreting booth. Your guests listen on Sennheiser receivers with headphones, handed out at the entrance, each language on its own radio channel and without an app. The sound comes straight from the mixing desk or the microphone, and our technician looks after the event.
A language model interprets for meaning and condenses when someone speaks fast. In blind tests with speeches from parliament and the EU and with talks, KADENZ conveyed 94 to 98 percent of the meaning, up from 79 to 92 percent before.
The language model and the voice each have a fallback: if the language model fails, direct translation takes over, and if a voice stalls, a backup voice steps in. If recognition drops out, it reconnects within seconds.
The voices come from ElevenLabs and sound like an interpreter at the microphone, with a female and a male voice for each language.
Recognition, translation and voice are separate building blocks. If a better service comes along, we swap it in. Before every new version, more than 160 automated checks run. In our own tests with talks, the translation followed the speech 4.5 to 7 seconds behind.
The software belongs to Eventforum, as do the transmitters and receivers. You get the hall, the technology and the interpreting from one provider, under one contract in Bern.
We considered data protection when building KADENZ. Your guests listen on our receivers, without an account or an app, and KADENZ processes no data about them. KADENZ does not clone voices and does not identify anyone by their voice. Error reports to the developers never contain audio, transcripts or names.
The recording with audio, transcript and glossary stays on our laptop on site for as long as agreed with you. Once that period has passed, KADENZ deletes it at the next start, with a record of the deletion. On request, we record nothing or delete the recording straight away.
For recognition, translation and voice, KADENZ uses specialised services. Which ones, in which countries they process data and for how long is set out in the data protection sheet under Swiss data protection law. On request, you receive it with the quote, together with a sentence to inform speakers and audience in advance. ElevenLabs, which provides the voices, is certified under the Swiss-U.S. Data Privacy Framework.
KADENZ is not intended for classified content; for that, your own team of interpreters works in our booths.
A language model interprets for meaning. In blind tests with speeches from parliament and the EU and with talks, KADENZ conveyed 94 to 98 percent of the meaning. With clear standard German and good sound, the translation is easy to follow. We add names and acronyms to the glossary in advance. For Swiss German, interruptions and confidential content, we recommend your own team of interpreters.
KADENZ interprets sentence by sentence: the translation starts as soon as a sentence is understood, about six seconds behind the speaker. In our own tests it was 4.5 to 7 seconds for talks. With very fast, dense speech the delay grows.
Up to four at once. We have tested English, French and Italian. If English is spoken on stage, KADENZ also interprets into German. Other languages on request.
No. We rent out booths, the system and receivers for your own team. KADENZ now also interprets with AI on request.
No. Your guests receive a receiver at the entrance and select the channel for their language. On request, they can also listen on their own phone, in the browser and without an app.
No. KADENZ runs on a laptop and sends each language on its own radio channel to the receivers. It therefore works in any room, from a workshop room to the event hall, and in several rooms at the same time. Your own team of interpreters works in our booths.
AI interpreting costs only a fraction of an interpreting team. You rent the interpreting system with booths and receivers as usual. We state the amount in your quote.
Yes, on request. After the event we send you the transcript for each language as a file. We keep the recording for as long as agreed.
KADENZ processes no data about your guests, does not clone voices and does not identify anyone by their voice. The recording stays on our laptop on site for as long as agreed. Recognition, translation and voice run through specialised services. Our data protection sheet, which we send with the quote on request, names the providers, countries and retention periods. KADENZ is not intended for classified content; for that, your own team of interpreters works in our booths.
Yes, if each person has their own microphone and people speak one after another. If several people speak at once, your own team of interpreters is better.
The language model and the voice each have a fallback: if the language model fails, direct translation takes over, and if a voice stalls, a backup voice steps in. If recognition drops out, it reconnects within seconds. Our technician watches every language in the studio and steps in when needed.
No. AI interpreting is an aid to understanding. What is said in the hall remains authoritative, for example for votes at a general meeting.
Yes. KADENZ runs at Eventforum at real events in the hall. Before use, we test every new version with a series of checks and a trial run of more than an hour. After the event, we ask for your feedback.
Yes. Bring a page from your programme. By arrangement, we run KADENZ in the hall with it and you listen to the translation on a receiver.
