Synthesia

Background

I get many, many questions about this popular technology. Synthesia.io is an AI video creation platform based in London. From my work in this space, and for my use-case, it is currently the world’s best avatar platform (Jan/2022). They have famously created avatars for David Beckham, Lionel Messi, and Snoop Dogg.

I use Synthesia for my Leta AI avatar and related avatars, featured on YouTube.

Each avatar is based on the face and lips of a real person (often a Synthesia.io employee or hired actor), and it is possible to create your own avatar for a fee. Note that the AI voice is a separate technology provided by Synthesia.io, and is not the original actor’s voice.

Creating the avatar

The actor stands in front of a green screen and is filmed in 4K resolution. They speak the following ~265-word script, and the process should be completed in about five minutes. (Closed Lips Image (💋) is the part of the script where the actor pauses and closes their lips):

💋
Hi, thanks for joining me. I am an A.I. presenter.
and today, I want to help you think positively.
💋
I’m sure you’ve heard before how important it is to have a positive frame of mind.
💋
Positive thinking is one of the greatest things we can do for ourselves.
It can help you believe in yourself, avoid worries,
and fill you with self-esteem and confidence.
💋
It can bring optimism, belief in your abilities and make you feel inspired.
It will give you the motivation to look for solutions and accomplish your goals.
💋
So look at the way you approach challenges ahead.
If you hear yourself think “I can’t do this, it will never get done on time.”.
Stop and think about a moment of success in the past.
Say to yourself, “I can get it done, I’ve made it work before and I will again”.
Make the choice to take a positive attitude.
💋
This will help you overcome your worries.
Value the optimism this brings and believe in yourself.
This will give you the motivation to look for solutions and accomplish your goals.
💋
The great news is that being positive can create benefits.
Not just for you but for those around you too.
The way that you think and the way that you act influences others
and it can be infectious.
💋
So use this to your advantage.
Make positive thinking work for you.
Feel inspired, optimistic, and motivated.
💋
When you believe in yourself you will find it easier to succeed and others will follow.
Judge for yourself, positive thinking really can be the key to success.
💋

Source: https://www.synthesia.io/get-your-own-synthesia-ai-presenter

From the spoken ~265-word script, Synthesia then run this video through software, and the avatar is output as a programmable head, and appears as an AI avatar in the Synthesia STUDIO.

Synthesia STUDIO

In the online Synthesia STUDIO, a text area allows the AI avatar to be fed with any text (there are sensitivity filters in place). This means that—just from the ~265 selected words in the script above—it can speak all ~1 million or so English words, as well as many other languages (see below).

In the case of Leta AI, I feed the verbatim responses from GPT-3 by pasting them into the online studio, and then exporting the end result to video, ready to ‘interact’ with it (see the ‘behind the scenes’ video).

Languages
Note that the AI voice is a separate technology provided by Synthesia.io, and is not the original actor’s voice. There are multiple language output options (about 80 languages and styles as of Jan/2022):

Arabic – Natural
Arabic – Default
Bengali – Original
Bulgarian – Natural
Catalan – Natural
Chinese (CN) – Natural
Chinese (HK) – Natural
Croatian – Natural
Czech – Natural
Danish – Natural
Danish – Original
Dutch – Natural
English (AU) – Natural
English (CA)
English (GB) – Narration
English (GB) – Natural
English (IE) – Natural
English (IN) – Natural
English (IN) – Original
English (NZ) – Natural
English (PH) – Natural
English (US) – Professional
English (US) – Upbeat
English (ZA) – Natural
Estonian – Natural
Filipino – Default
Filipino – Original
Finnish – Natural
French (CA) – Natural
French (CA) – Original
French (CH) – Natural
French (FR) – Natural
German (AT) – Natural
German (CH) – Natural
German (CH) – Natural Deep
German (DE) – Natural
Greek – Natural
Gujarati (IN) – Natural
Hebrew – Natural
Hindi – Default
Hindi – Natural
Hungarian – Natural
Indonesian – Default
Indonesian – Natural
Irish – Natural
Italian – Natural
Japanese – Natural
Korean – Natural
Latvian – Natural
Malay – Natural
Marathi (IN) – Natural
Norwegian – Default
Norwegian – Natural
Polish – Natural
Portuguese (BR) – Natural
Portuguese (PT) – Default
Portuguese (PT) – Natural
Romanian – Natural
Russian – Natural
Serbian – Original
Slovak – Natural
Slovenian – Natural
Spanish (AR) – Natural
Spanish (CO) – Natural
Spanish (ES) – Natural
Spanish (LA) – Original
Spanish (MX) – Natural
Swedish – Natural
Tamil (IN) – Natural
Tamil (IN) – Original
Telugu (IN) – Natural
Thai – Natural
Turkish – Default
Turkish – Natural
Ukrainian – Default
Urdu (PK) – Natural
Vietnamese – Default
Vietnamese – Natural
Welsh – Natural

Dr Alan D. Thompson is an AI expert and consultant, advising Fortune 500s and governments on post-2020 large language models. His work on artificial intelligence has been featured at NYU, with Microsoft AI and Google AI teams, at the University of Oxford’s 2021 debate on AI Ethics, and in the Leta AI (GPT-3) experiments viewed more than 5 million times. A contributor to the fields of human intelligence and peak performance, he has held positions as chairman for Mensa International, consultant to GE and Warner Bros, and memberships with the IEEE and IET. Technical highlights.