InterpretBank ASR

Automatic Speech Recognition for Interpreters

Start live demo

(Live demo requires Google Chrome. A preloaded EN-FR glossary will be used.)

Integrating Automatic Speech Recognition (ASR) in the interpreter's workstation

InterpretBank ASR is an experimental feature of InterpretBank that automatically looks up your glossary and numbers while you are interpreting. This tool has been developed by Dr. Claudio Fantinuoli at the university of Mainz/Germersheim.

Video showing InterpretBank ASR for numbers

The Project

What it does

InterpretBank is the world leading computer-assisted interpreting tool. We are now adding AI feature to the interpreter workstation. Terminology and numbers are challanging for interpreters. We are invastigating the technical and cognitive feasablity to use ASR to support interpreters in these areas.

Terminology

Automatic lookup of specilized terms

Numbers

Automatic prompt of numbers

Live

InterpretBank ASR listens to the speaker and makes you suggestions

Quick

...everything happens in real time

Hardware requirements

In a typical setting, the audio channel (what you hear in your headphones while interpreting) needs to be sent to the computer. To do so, you have several options:

- many booth consoles (such as the Barinas TC-F06/16) have 2 audio lines-out (3.5mm jacks). One will be used for your headphones, the second will be used to connect the console to the audio-in jack of your computer.

- if the interpreting console has only 1 audio line-out, you may use an audio-splitter to have 2 audio lines-out (one for your headphone, one for the computer line-in)

- on your computer: if your computer does not have a line-in (most laptops have not), you need an external USB sound card with line-in (see picture). Typical price between 10 and 30 Euros. Note that there is a difference between a microphone line-in (in the picture below in green) and an audio line-in (in the picture below the white and red chinch connectors). To link interpreting console and your computer you need a sound card with audio line-in. You will also need an audio cable to connect the two, in the setting described below a cable 3,5 jack to chinch (see picture). Note: as speech recognition is sensible to audio quality, we suggest you to buy a good quality external sound card.

Research

Research around CAI tools, ASR and interpreting

InterpretBank has been created by Dr. Claudio Fantinuoli (PhD) at the university of Mainz/Germersheim. For a list of publications related to CAI tools and Automatic Speech Recognition, visit this page.

Contact

InterpretBank ASR is based on an original idea by Dr. Claudio Fantinuoli, University of Mainz, Germany. If you need more info about Claudio's work, here is his Academic page.