LoginSign Up
← Back to all sparks
S

Synthesia

AI-ASSISTANTS
Velocity0.0

AI video generation platform with realistic avatars and voiceovers for training and marketing.

www.synthesia.io ↗

Synthesia is becoming a general AI video editor — avatars are now one feature, not the product.

video-generationavatar-evolutioncontent-ingestionthird-party-modelsrelease-slowdown
◆Current state
Synthesia has spent the last six months extending its product surface well beyond AI avatar generation. The Editor now ingests external screen recordings (MP4 → transcribed, scene-split, editable Synthesia video), accepts .pptx with speaker notes as voiceover, and runs an AI Playground that exposes third-party models — Sora 2, Veo 3.1, FLUX.2, Nanobanana Pro — directly inside the canvas. Avatar capability also broadened: action-taking stock avatars with arbitrary backgrounds, speech regeneration, and per-voice speed control. The release cadence has slowed visibly since March, with no public updates in the past two months.
◆Where it's heading
The strategic move is from 'create a video by typing a script for an avatar' to 'turn any input (slides, recordings, prompts) into a Synthesia-editable video,' with third-party genAI models embedded in the canvas. Avatars are repositioning as one input among many, not the headline. The pause in release cadence since March is notable for a product that was shipping every two to three weeks through Q4 2025 — could indicate a larger release in flight, a strategic reorientation, or commercial pressure squeezing the public-facing tempo.
◆Prediction
The next visible release will likely be the next-generation avatar tier (the action-taking stock avatars were called 'one of the most exciting updates of the year' in November, so an upgrade or open-prompt avatar variant is overdue), or a foundational change to the ingestion pipeline that ties the screen-recording and PowerPoint surfaces into a single 'video from anything' flow. If the silence continues past Q2, that's a signal worth watching.

◆Recent moves

  1. 2mo ago

    📹 Turn External Screen Recordings into editable Synthesia videos

    ⚡ SPARK

    External MP4 screen recordings now upload directly into the Editor, with automatic transcription, filler-word removal, and scene splitting. Synthesia's ingestion surface expands from synthetic content to existing user-recorded media — meaningful shift in what the product is for.

    View source ↗
  2. 3mo ago

    Meet the new PowerPoint to Video 🆕🎥

    Updated PowerPoint to Video flow: upload a .pptx and get an editable Synthesia project, with speaker notes converted to voiceover. Existing feature significantly modernized; fits the broader 'turn any input into a Synthesia video' arc.

    View source ↗
  3. 4mo ago

    🔁 Get the voice right: regenerate speech and adjust voice speed

    Speech regeneration and voice-speed control finally land — a long-requested pair of TTS controls. Closes a usability gap that had pushed users to either accept the default take or redo paragraphs from scratch.

    View source ↗
  4. 5mo ago

    🧪 Experiment with media creation in AI Playground

    AI Playground tab surfaces third-party genAI models inside the Editor sidebar for ad-hoc media creation. Positions Synthesia as a host for a curated set of models rather than a single in-house generator.

    View source ↗
  5. 5mo ago

    Create any images with FLUX.2 & Nanobanana Pro 🍌

    Image generation lands inside the Editor via FLUX.2 and Nanobanana Pro. Removes a context switch — users no longer need to leave Synthesia to source generated images for their projects.

    View source ↗
  6. 6mo ago

    New Avatars that can take action 🧑‍🎤

    ⚡ SPARK

    Stock avatars can now take action in any background and outfit, from a single prompt. Synthesia framed this as one of the year's most significant updates — first time the action surface is exposed at this control depth on stock avatars.

    View source ↗