InterpretBank ASR is a virtual assistant that supports you during interpreting by transcribing speech in real time. While you interpret, it displays the spoken text, allows you to translate the transcription on the fly, highlights numbers, and gives you instant access to terminology from your glossaries.

InterpretBank ASR is designed for high precision in real-time speech recognition and can be used both in simultaneous interpreting (as a Digital Boothmate) and in dialogic interpreting.

Two versions are available

• InterpretBankASR Cloud: browser-based session hosted on the InterpretBank ASR server. Best for collaboration and quick access from any computer, with broader language availability and the highest transcription precision.
• InterpretBankASR Offline: local desktop app using downloaded speech models on your machine. Best when you need local processing with no internet dependency and max. data confidentiality.

Main differences

• Where transcription runs: InterpretBankASR Cloud = remote ASR engine, InterpretBankASR Offline = local ASR engine on your computer.
• Sharing: InterpretBankASR Cloud supports share links for colleagues; InterpretBankASR Offline is local-only.
• Internet dependency: InterpretBankASR Cloud needs internet during use. InterpretBankASR Offline can transcribe locally once configured, but still needs internet for license validation and model download/update checks. No audio data leaves the computer.
• Glossary source: Both versions use glossaries from your main InterpretBank app. In InterpretBankASR Cloud, glossary data is uploaded for the session to a GDPR-compliant server. In InterpretBankASR Offline, glossary data is accessed directly on your computer.

Transcription engines, language availability, and privacy for InterpretBankASR Cloud

• STANDARD transcription engine: unlimited use included in subscription (subject to availability). Not GDPR-compliant.
• PREMIUM transcription engine: GDPR-compliant, usage-based with credits. Best recognition precision across all languages. To purchase credit for the cloud PREMIUM engine, click here.
• Language support: speech recognition supports Albanian, Arabic, Armenian, Bulgarian, Catalan, Chinese, Croatian, Czech, Dutch, English, Finnish, French, German, Greek, Hungarian, Italian, Japanese, Korean, Norwegian, Polish, Portuguese, Russian, Slovenian, Spanish, Turkish, Vietnamese. Machine translation is available into virtually all major target languages supported by the session.

Transcription engines, language availability, and privacy for InterpretBankASR Offline

• InterpretBankASR Offline downloaded transcription engine: unlimited use included in subscription.
• Privacy: transcription runs locally on your computer.
• Language support: depends on which offline language resources are installed in Settings > Model downloads. Only installed languages are available for transcription.

Note that lookup of glossary inflected forms (singular, plural, etc.) is currently optimized for English, French, and Spanish.

InterpretBankASR Cloud: overview

The cloud flow runs in your browser and is built around a live session URL. In the backend, the session owner opens /live/<sessionId>, and viewers can join with a share link /share/<shareId>.

Use this version when you need collaboration, quick start, and access from different devices.

InterpretBankASR Cloud: create a session without glossary

This type of session lets you use speech recognition and machine translation without uploading or activating any glossary.

1. In InterpretBank Desktop, open the menu AI Tools | Open InterpretBank ASR | Without any glossary, or
2. In the WebApp, select the Start ASR session icon in the sidebar.
3. A webpage will open in your web browser. If Chrome is not your default browser, copy and paste the URL into Chrome.
4. This session provides live transcription and machine translation.

InterpretBankASR Cloud: create a session with a glossary

This type of session lets you use speech recognition, machine translation, and automatic suggestions based on your glossary. To enable this, the glossary must be uploaded to the cloud.

1. Open a glossary in InterpretBank Desktop.
2. In InterpretBank Desktop, select the menu AI Tools | Open InterpretBank ASR | With current glossary, or
3. In the WebApp, open the actions menu for a glossary and select Lookup with ASR.
4. A webpage will open in your web browser. If Chrome is not your default browser, copy and paste the URL into Chrome.
5. This session provides live transcription, machine translation, automatic glossary lookup, and highlighting of numbers and names.

InterpretBankASR Cloud: share a running session with other interpreters

This feature is available only for sessions created with a glossary.

You can share the session output on your screen with colleagues, without requiring them to have InterpretBank installed or to connect to the event’s audio.

To share a session, click the Settings icon in the top bar. A pop-up window will appear, showing a URL that you can share via email or other channels.

When opening the link, recipients can choose their preferred target language.

InterpretBankASR Offline: setup and run a session

The offline version is a local desktop app. It transcribes with locally installed speech-recognition resources and accesses glossary data directly on your computer.

1. Install and open InterpretBankASR Offline.
2. Wait for license validation at startup. The Start button stays disabled until license and model checks pass.
3. Open Settings > Model downloads and download at least one model for your source language (small/medium/big).
4. In Settings > Model and Audio, choose the model and input audio device.
5. Open Glossary selector, choose glossary and subglossary, then click Load glossary.
6. Back in Live, set Source and Target language.
7. Click Start to begin transcription. Use Clear to reset transcript and terminology.

Important: this version does not generate share links. It is intended for local usage on a single workstation.

How to connect the audio of a physical console (on-site)

To use InterpretBank ASR, you must send the audio you hear in your headphones (the interpreter feed) from the physical console to the app. This is super easy. To connect a Mac or PC to an interpreting console, you will need the following cables and adapters:

• Male-to-male 3.5 mm audio AUX cable (example)
• External USB sound card with audio input: a simple mic-in (example) or a sound card with a proper line-in (example)

Connect the line-out of the booth console to the audio-in of the external sound card using the AUX cable. Then connect the sound card to your computer via USB.

Note: using a sound card with a proper line-in is recommended. If you use a sound card with a mic-in, keep the input volume low, as mic inputs are more sensitive than line inputs and may cause distortion.

Once the line-in is connected to the sound card and the sound card is plugged into your computer, simply start an ASR session. By default, InterpretBank ASR listens to your sound card input, so no additional configuration is usually required.

When prompted by Chrome, make sure to grant permission to access audio input. If you need to change the default audio source, open Settings > Audio in InterpretBank ASR and select the desired input from the dropdown menu.

Troubleshooting. If no transcription appears, but the audio meter is moving in Settings > Audio, the issue is often related to Chrome audio permissions. In this case, resetting Chrome’s audio permissions usually solves the problem.

To reset audio permissions in Chrome:

• Open the ASR session page in Chrome.
• Click the lock icon (or settings icon) to the left of the address bar.
• Set Microphone to Ask or Allow.
• Reload the page.
• When prompted again, allow Chrome to access the audio input.

After reloading, verify that the correct audio input is selected in Settings > Audio and that the audio meter is active. Transcription should start within a few seconds.

How to connect a virtual audio source (Zoom, etc.) on Windows

This configuration allows InterpretBank ASR to listen to a source on your computer (youtube, zoom, etc.) and route the audio also to InterpretBank ASR. Note that you can not speak in your mike (for example for RSI) at the same time. See the limitation sessions below and how to configure your working station for RSI.

You can use the Windows build-in virtual sound card called Stereo Mix (should you not see Stereo Mix among the sound cards, refer to this short video on how to show it):

• open the Sound options of Win10 or Win11 by right clicking on the speaker icon in the menu bar, and selecting "Sound", see screenshot 1
• make sure that in tab "Playback" the Speaker (or headphones) are selected, see screenshot 2
• make sure that in tab "Recording" the Stereo Mix is selected, see screenshot 3

Troubleshooting

Modifying standard audio on a Windows machine can be daunting. If the above solution does not work, you might want to check these:

• Stereo Mix on Windows 11

How to connect a virtual audio source (Zoom, etc.) on macOS

This configuration allows InterpretBank ASR to listen to a source on your computer (youtube, zoom, etc.) and route the audio also to InterpretBank ASR. Note that you can not speak in your mike (for example for RSI) at the same time. See the limitation sessions below and how to configure your working station for RSI.

You can use the VB-Cable virtual sound card. You can download it here. Once installed, VB-Cable will appear in the list of your sound devices. You can see your sound devices in macOS System Preferences | Sounds.

• Navigate to System Preferences | Sounds and select VB-Cable both in tab Output and Input. Now the sound on your computer will be automatically routed to the soundcard. You will not hear any audio from the speaker/earplugs anymore. InterpretBank ASR will be able to hear the sound in your computer.
• Open InterpretBank ASR and select VB-Cable as input device in the audio settings. If prompted, allow microphone/audio input permissions.

Note on Remote Simultaneous Interpretation (RSI)

For Remote Simultaneous Interpretation (on platforms like Zoom, i.e. whenever you normally use one device to listen and speak), we recommend using two separate computers – one dedicated to the videoconferencing platform (where you listen and translate) and the other for InterpretBank ASR. Configuring a single computer's audio channels to accommodate both applications can be complex and is generally not advised.