LEXI AI-Powered Live Automatic Captions

Our LEXI automatic captions leverage the latest AI to deliver an average accuracy of 98.7% – which rivals the quality achieved by human captions, but at a fraction of the cost.  

Yes, LEXI automatic captions work with our entire suite of encoders. This includes: 

• Alta: A virtualized, API-powered IP video encoder that saves time and reduces costs with next-gen workflows.
   
• Encode Pro: An HD-SDI encoder trusted by major TV networks worldwide, offering low-cost, low-latency caption display.
   
• Encode 4K: A native 12G encoder that allows you to fLEXIbly work with broadcast, live stream and in-room spaces while encoding in 4K or HD-SDI. 

• Falcon: A budget-friendly cloud encoder that makes it easy to add captions to RTMP/S streams.  

Topic models (sometimes also known as custom dictionaries) are databases of specific terms and phrases that teach an Automatic Speech Recognition (ASR) engine, so it produces the words correctly when it ‘hears’ them. With LEXI, you have the option to add your own topic models. This process makes the live captions more accurate when consumers receive them to their screens. 

Regulatory guidelines for live captioning vary greatly between countries and regions. While LEXI meets regulatory requirements for quality and time delay in many regions, some countries have specific requirements that LEXI does not yet meet. If you are unsure of your country’s requirements, please get in touch for more information. 

No, LEXI is its own subscription fee.

LEXI works best in situations with clear audio signals, like in broadcast. If the session was a live broadcast, you can try to add more detailed entries to your Topic Model. You may also want to consider upgrading to our Smart LEXI service, where our trained professionals create the Topic Models for you.

When creating a Topic Model, you want to include any jargon that you think may appear in your session. You want to include proper nouns like names of people and places, for example. Also, it’s generally useful to include variations on a given term. For example: Alan Grant, Dr. Grant, etc.

The current limit is around 1500 entries, but this will increase with the release of an updated Topic Model storage system. The likelihood of a given term being recognized doesn’t necessarily correspond with the size of the Topic Model.

You can cancel LEXI by logging into your EEGCloud.tv account, accessing your Billing Dashboard and clicking on LEXI, and then the ‘cancel subscription’ button. You may also reach out to a salesperson to discontinue service.

When you purchase a monthly subscription and utilize all of your subscription hours, your LEXI subscription does not automatically cancel. Should you go over your monthly allotted hours, you will be billed for each hour used over and above your monthly subscription. The per-hour rate depends on your LEXI subscription. The larger your monthly subscription, the cheaper your LEXI additional hours are!

Our LEXI Local solution couples with your existing EEG equipment to give you an unlimited ASR use system. This uses a 1-time annual subscription fee with no additional yearly costs.

• English (Australian)*
• English (UK)*
• English (US)*
• Global Spanish*
• Catalan
• Basque
• Galician
• Arabic
• Mandarin
• Czech
• Dutch
• French
• German*
• Hindi
• Italian*
• Japanese
• Korean
• Portuguese*
• Bashkir
• Belarusian
• Bulgarian
• Croatian
• Danish
• Esperanto
• Estonian
• Finnish
• Greek
• Hungarian
• Interlingua
• Latvian
• Lithuanian
• Norwegian
• Polish
• Romanian
• Russian
• Slovakian
• Slovenian
• Swedish
• Turkish
• Ukrainian
• Welsh
• Indonesian
• Malay
• Mongolian
• Thai
• Vietnamese
• Cantonese
• Uyghur

The most common use case would be to prevent overlap with static graphics that would be to the left of the closed caption box, but depending on the end player, that might just produce a blank space without closed captions, but still overlapping on graphics.

No, stereo audio does not qualify as multi-track.

Yes. In your EEG Cloud Account, navigate to LEXI Live and find the LEXI Instance you want to add the Topic Model to. From there, click on Settings, and you can then add the appropriate Topic Model from the Topic Model dropdown.

The Vocabulary Control allows you to add custom words & phrases to LEXI’s vocabulary. Common entries include proper nouns like names of people and places, terms you think LEXI may have difficulty processing, and any session-specific vocabulary.

The Substitutions component functions to recognize words or phrases and then either blacklists or reformats them. It kind of acts like a ‘find and replace’ tool in real-time captioning. It can be used to ensure spelling standards, proper formatting and prevent unwanted misrecognition errors.

CCMatch is an encoder module that is only available on specific products (Encoder Pro, Encoder 4K, or Alta). Any encoder with that module enabled can sync the captions better to the audio/video, whether they come from LEXI or a Live captioner.

Yes, in LEXI’s Advanced Options, you must turn on LEXI Vision. LEXI Vision uses AI to set captioning screen position automatically based on your video content, avoiding text and faces. LEXI Vision will override your fixed row position settings.

Smart LEXI is full-service. Our professionals build the Topic Models for you with their expertise and experience.

3 days in advance would be great, but the earlier the better. If our team has prep material for your Smart LEXI session days in advance, they’ll be able to begin curating the Topic Model best suited for that session.