Synthesia

Synthesia is a British artificial intelligence company that operates a browser-based platform for creating presenter-led video from text scripts. Users type a script, choose a digital avatar from a library of realistic synthetic presenters, and generate a finished video in minutes — without cameras, studios or actors. The platform supports more than 140 languages and is used primarily by enterprises for training, onboarding and internal communications.

History

Synthesia was founded in London in 2017 by Victor Riparbelli, Steffen Tjerrild, Lourdes Agapito (a professor at University College London) and Matthias Niessner (a professor at the Technical University of Munich), emerging from academic research in computer graphics and neural rendering.^[1] The company raised a US$90 million Series C in 2023 led by Accel with NVIDIA's NVentures, reaching a US$1 billion valuation, followed by a US$180 million Series D in January 2025 at US$2.1 billion led by NEA.^[2]

Synthesia crossed US$100 million in annual recurring revenue in April 2025, and market tracker Sacra estimated ARR of roughly US$146 million by September 2025 as revenue grew about 80% year on year.^[3] In January 2026 the company closed a US$200 million Series E led by Google Ventures with NVIDIA and existing investors participating, valuing the company at US$4 billion.^[2] The platform released Synthesia 3.0 in October 2025, introducing interactive video agents that can answer viewer questions inside a video and run role-play or comprehension checks, built on the EXPRESS-2 avatar model.^[3][4]

Key Concepts and Technology

Synthesia's pipeline converts a text script into speech with text-to-speech synthesis, then animates an avatar's lip movements, facial expressions and upper-body gestures to match the audio — a problem that requires frame-accurate alignment of phonemes with facial muscle dynamics. Key elements of the platform include:

• Stock avatar library: more than 240 pre-made presenters covering a range of ages, ethnicities and styles.
• Custom avatars: digital twins created from a few minutes of smartphone footage or a professional shoot, with voice cloning attached.
• One-click translation: an original video can be re-rendered in dozens of languages with re-synchronised lip sync; the company reports around 40% of videos created are translated versions of an original.^[3]
• Enterprise controls: single sign-on, SCORM export for learning management systems, and compliance certifications including SOC 2 and ISO/IEC 42001 — the practical features that drive corporate procurement.^[3]

Competitors include HeyGen, D-ID, Colossyan and Hour One, with HeyGen the closest challenger on avatar realism and shipping speed.

Applications or Impact

Synthesia's main use cases are corporate learning and development, employee onboarding, compliance training, product marketing and customer education — scenarios where a studio shoot per language or per script update would be prohibitively slow. By collapsing video production from days to minutes, the platform shifted video from a production budget line to a self-serve tool for HR and communications teams. It also intensified debate over synthetic media: custom avatars require consent from the depicted person, and organisations increasingly label AI-generated presenters to maintain trust.

>See Also

• AI Video Generation
• HeyGen
• Text-to-Speech
• Voice Cloning
• ElevenLabs
• Deepfake