AI Image, Video and Voice

Oct 9

Zack London, known on YouTube as Gossip Goblin, made an AI feature that the filmmakers say is the first AI film to get an MPA rating. It is rated R, moves to December 4, and will go for Best Animated Feature. Here is what trade press confirms and what rests on the producers’ word.

Oct 9

Sperid Labs released Iris-3B on October 8, 2026, a 3B text-to-image diffusion transformer that outputs every pixel directly, with no VAE and no latent space. The paper reports the model matches Qwen-Image on OneIG. It also reports that pixel space did not beat latent models on depth or restoration.

Oct 9

Voyager launched on October 8, 2026 as an "open harness" for creative work. The founder calls it the Codex for creative work. We read the launch thread, the FAQ and the pricing page. One surprise: the app is not open source.

Oct 6

Kandinsky Lab open-sourced Kandinsky 6.0 Video on October 6, 2026: two diffusion models that generate five-second clips with synchronized 44 kHz audio, released with code and weights under the MIT license. This post covers the architecture, the claimed results, the GPU requirements, and where it fits next to other open video models.

Sep 29

On September 24, 2026, Dev Mode Discord screenshots showed VIDEO 4.0 Preview and VIDEO 4.0 Flash Preview inside Kling's creator app, with a duration slider from 3 to 30 seconds. Kuaishou has not announced the model, published a model card, or listed 4.0 in the API changelog. This news post separates leaked UI from what you can ship on today.

Sep 26

Sep 26

Sep 24

Fish Audio introduced Drama 3, a preview text-to-speech model it calls the most controllable ever: describe tone and character in simple language, change voice mid-sentence, render multi-character scenes and regenerate a single word. Access is gated, pricing is unpublished. Here is what is confirmed, what to test, and how it compares.

Sep 20

A viral blog post and its 750-comment Hacker News argument prove that "AI slop" posters aren't a model limitation — they're what happens when nobody tells the model which design language to use. The same fix applies to AI-generated code and prose.

Sep 17

ElevenLabs announced Reception on September 16, 2026, and the post drew 1.2 million views. The most useful thing in the thread was not the demo. It was small business owners asking the three questions the announcement did not answer: what happens off-script, what happens on interruption, and what happens to trust.

Sep 15

Polymarket's X account shared a demo of a paralyzed woman with severely impaired speech holding real-time conversation through a brain implant and an AI-generated voice. The "mind reading" framing is wrong — this is residual speech-signal decoding plus personalized voice synthesis. Here's the corrected explanation and the applied-ML problem underneath it.

Sep 10

Suno launched v6, a new generation of its AI music generation models, built on licensing agreements with Warner Music Group and BMG rather than the unlicensed training data approach that triggered lawsuits from major labels. explainx.ai covers what changed in Suno's approach, why the labels settled rather than continued litigating, and what it means for the broader AI music generation landscape.

Sep 10

A single builder trained a computer vision system that tracks their own tennis strokes — ball speed, forehand vs. backhand, shot placement, and body position at contact — from plain iPhone video. The pipeline pairs Roboflow Agent for auto-labeling and fine-tuning with Claude Code driving the terminal, plus MediaPipe for pose estimation. explainx.ai breaks down the exact stack and why this kind of project is now a weekend build instead of a multi-week one.

Sep 9

HyperFrames is HeyGen's open-source framework for turning plain HTML into frame-accurate MP4 video, shipped with 20 agent skills for Claude Code, Cursor, Gemini CLI, and Codex. This guide covers the composition model, the skills-router architecture, the explicit Remotion comparison, and what you can actually build with it today.

Sep 7

VoiceStudio (formerly OmniVoice Studio) is an open-source, AGPL-3.0 desktop app that clones voices, dubs video into other languages, dictates system-wide, and produces audiobooks — all running locally, with 16 TTS engines and 11 ASR engines to choose from.

Sep 3

Ato announced its second-generation device on September 2 after more than 2,500 older adults used it in beta over a year. It is voice-only, has no camera, and positions itself as a coordination layer between relatives, caregivers and providers. The post did 2.9 million views — and the top reply was somebody telling the audience to go visit their parents instead.

Sep 2

Developer Rehan Sheikh turned faster-than-playback H3 Max generations into an always-on “interdimensional cable” stream influenced by viewer prompts. The demo proves continuous generative television is technically possible—and exposes brutal economics, weak memory, moderation, and copyright problems.

Sep 2

Google Workspace announced Google Pics on September 1, 2026 — an AI image creation and editing app built on Nano Banana 2 that lets teams edit individual objects, translate in-image text, and collaborate on images the way they already do on Docs and Slides. Here's what it actually does, why commenters immediately asked how it differs from Nano Banana and Imagen, and whether it's a real threat to Canva.

Sep 2

Sep 1

A KAIST-led team with NUS and Singapore Management University built SweepLED — a $7 LED device that clips onto a smartphone and uses AI to spot hidden cameras in under 5 seconds with 94% accuracy. It works by analyzing how a lens reflects light from multiple angles, not by hunting for a single bright spot, so it catches cameras whether they're powered on or off.

Aug 31

Lukas Kuhn et al. posted LeVJEPA (arXiv:2608.27395): video representation learning without EMA targets, stop-gradients, or pixel decoders. At ViT-S it uses up to 20.8× less compute than V-JEPA 2 at matched epochs — relevant for robotics and world-model builders.

Aug 31

Motion (motion.so) from Mosaic (YC W25) is now generally available — an AI agent for motion design that accepts product links, X threads, YouTube videos, and DESIGN.md context to storyboard and render launch films you can edit in place.

Aug 31

On August 30, 2026, OpenShot 4.0 shipped the biggest workflow upgrade in the project's history: a dedicated Color View with scopes and LUTs, a Recording View for screen/webcam/mic capture, 10 new effects, and local AI object masks via ONNX — no subscription. A 319-point Hacker News thread compared it to DaVinci Resolve, Kdenlive, and LosslessCut. explainx.ai breaks down what builders get from a free GPLv3 editor that runs YOLO on your machine.

Aug 29

Aug 29

On August 28, 2026, Diffusion HQ (YC F24) open-sourced a video editor built on one idea: every edit is code, not an opaque render. The pitch is "code is the new database" — an agent can read, diff, and re-run a timeline the way it works a codebase. explainx.ai looks at the manual-edit-to-reusable-skill workflow, how it compares to ViMax and OpenCut, and whether editing-as-code actually fixes agent context loss.

Aug 29

MiniMax announced Fast H3 v1 around August 29, 2026 — a faster inference variant of the H3 video model that the company says hits roughly a 14x speedup on NVIDIA Blackwell, aimed at real-time and faster-than-real-time open video generation. Details are thin. explainx.ai covers what real-time video unlocks for builders, how Fast H3 sits next to H3 Max and H3C, and the caveats that come with a provider-reported number.

Aug 28

Aug 28

fal Research shipped H3 Max on August 26 — a post-trained MiniMax H3 that renders a 5-second 768p clip with synced audio in under three seconds. Ethan Mollick called it a line being crossed: generation now takes less time than watching the result. explainx.ai covers the benchmarks, the $0.08/second pricing, and what breaks when video generation becomes interactive.

Aug 28

Daily, the team behind the open-source Pipecat voice-AI framework, released PhoneLLM Alpha 1 on August 27, 2026 — an open-weights fine-tune of NVIDIA Nemotron 3 Nano claiming GPT-5.6 Terra-level quality on phone-support tasks at roughly 1/18th the cost and faster first-token latency. Here is what it actually claims, how the numbers stack up, and why this is an early alpha, not an independently verified benchmark.

Aug 27

On August 27, 2026, UCLH announced that a 48-year-old man became the first person in the world to have a brain tumour removed while an AI system analysed the live surgical video feed. The viral version of this story implies a robot did the operation. It did not. This is what the system actually does, why "live" is the hardest word in the sentence, and who really audits the code.

Aug 26

Aug 25

On August 24, 2026, Prime Minister Andy Burnham and President Volodymyr Zelenskyy signed a UK-Ukraine AI partnership giving Britain first foreign access to Avengers AI Labs — a platform built on five million annotated battlefield frames from DELTA sensors and drone feeds. explainx.ai breaks down what the dataset actually contains, what UK teams can build with it, and what the declaration does not legally bind either side to.

Aug 22

Nari Labs, the team behind the open TTS model Dia, published a technical breakdown of how they pushed Qwen3-TTS to 10 requests/second and sub-50ms time-to-first-audio on a single H100 — at roughly $2 per million characters. explainx.ai walks through the five serving techniques and the Hacker News practitioner Q&A that followed.

Aug 19

BGRemover.video is an AI tool that strips or replaces a video's background without a green screen, exporting transparent WebM/MP4 in three steps. We cover how the AI works, pricing tiers, batch processing, and an honest note on what re-encoding does to embedded metadata like C2PA content credentials.

Aug 18

Three months after Sonic-3.5, Cartesia released Sonic-3.6 in beta. It leads Artificial Analysis on both provider-voice and controlled-voice streaming leaderboards. Hear the English and Hinglish demos, then read what the Elo numbers actually mean for a production voice stack.

Aug 18

Roboflow ML engineer Piotr Skalski published a VLM benchmark showing GPT-5.6 Sol is a massive leap for OpenAI on object detection and counting — up from 13.8 to 46.2 mAP@50 — but Gemini 3.5 Flash still beats it on most vision tasks at roughly a third of the cost. The post hit #1 on Hacker News twice, and Skalski himself now says Gemini 3.7 Flash is the better pick.

Aug 18

Stability AI released two new ways to work with Stable Audio 3.0 on August 18, 2026 — a DAW plugin that puts generation directly inside Ableton and Logic, and a rebuilt StableAudio.com built for iterating on a track rather than generating once and leaving. Both run on Stability's commercially safe, fully licensed models. Here's what shipped, how licensing actually works, and how it compares to open alternatives.

Aug 9

Higgsfield launched a limited-time offer on August 7, 2026: unlimited generations on ByteDance's Seedance 2.5, no per-clip credit deduction, for up to 33 days depending on your plan. The catch is the fine print — resolution is capped at 720p, clip length varies by tier, and you have to manually flip an "Unlimited" toggle or it silently burns your credits anyway.

Aug 5

On August 4, 2026, phone-agent company Bland launched Speech v3, a standalone voice model it calls the "world's first Human Speech Engine." The centerpiece is a case study restoring a stroke survivor's voice — here's what the benchmark claim actually rests on and what the launch means for Bland's business.

Aug 1

RF-DETR is a real-time detection transformer from Roboflow built on a DINOv2 backbone, spanning Nano to 2XLarge across detection, segmentation, and keypoint tasks. It hit ICLR 2026, and Roboflow now runs its architecture search directly on the platform. Here's what it is, how it benchmarks, and how to run it.

Jul 31

Jul 31

Jul 30

Jul 29

Jul 26

Jul 23

Jul 19

Jul 19

Jul 15

Jul 10

Jul 9

Jul 8

Jul 8

Jul 7

Jul 4

Jul 3

Jun 27

Jun 27

Jun 25

Jun 23

Jun 23

Jun 22

Jun 21

Jun 21

Jun 20

Jun 19

Jun 18

Jun 17

Jun 16

Jun 13

Jun 4

May 31

May 27

May 26

May 24

May 22

May 21

May 20

May 4

Apr 25

Apr 13