Vannarot Roeung
Recent posts by Vannarot Roeung
Darwin-TTS Adds Emotion to AI Voice With No Training Required
Researchers published Darwin-TTS on April 15, 2026, a text-to-speech model that adds emotional expression to AI voice without any training, fine-tuning, or new data.
Google Gemini 3.1 Flash TTS Beats ElevenLabs in Quality Benchmarks
Google released Gemini 3.1 Flash TTS on April 15, 2026, a text-to-speech model that outperforms ElevenLabs v3 in quality benchmarks while offering a generous free tier.
Splice Launches AI Sample Variations with Built-In Creator Payouts
Splice launched Variations and Craft on April 15, 2026, letting producers remix any sample from its 3-million-sound library while automatically compensating the original creator.
Shutterstock Launches Commercial AI Video Generator
Shutterstock launched its AI Video Generator on April 15, combining text-to-video and image-to-video models from Google and Runway with commercial-ready licensing for brand and agency workflows.
Adobe's Firefly AI Assistant Automates Creative Workflows
Adobe launched the Firefly AI Assistant on April 15, 2026, a creative agent that takes a plain-English brief and executes multi-step workflows across multiple Creative Cloud apps at once.
Adobe Firefly Video Editor Adds Kling 3.0 and Frame.io Drive
Adobe Firefly Video Editor adds Kling 3.0 AI video models, Frame.io Drive, and Enhance Speech at NAB 2026, giving creators a full AI production pipeline.
Midjourney V8.1 Alpha Ships 3x Faster HD Generation
Midjourney V8.1 Alpha brings 3x faster HD mode at 3x lower cost, 50% faster standard generation, restored image prompts, and stable moodboard references.
ComfyUI Adds Frame-Synced AI Music via Sonilo
ComfyUI now supports Sonilo via Partner Nodes, letting creators generate full-length soundtracks that sync to video footage frame by frame.
Microsoft MAI-Image-2-Efficient Cuts AI Image Cost 41%
Microsoft launched MAI-Image-2-Efficient on April 14, 2026, a distilled variant of its flagship image model with 41 percent cheaper output tokens and 22 percent faster generation on H100 hardware.
Meshy and Formlabs Open the Text-to-3D Loop
Meshy AI and Formlabs Form Now ship the first end-to-end text-to-physical 3D pipeline. Under five minutes from prompt to order, 48-hour delivery, 97% slicer pass rate. The moment AI 3D collapsed the screen-to-studio gap.
Nucleus-Image Is the First Open-Source MoE Diffusion Model
NucleusAI released Nucleus-Image on April 14, a 17-billion parameter diffusion model that activates only 2 billion parameters per image -- cutting compute without sacrificing quality.
HeyGen Launches Developer Platform for AI Video Agents
HeyGen launched a developer platform with an open-source CLI, an HTML-to-video rendering framework called Hyperframes, and agent skills for Claude Code, Codex, and Cursor.
DaVinci Resolve 21: Free AI Tools Challenge Adobe
Blackmagic Design announced DaVinci Resolve 21 at NAB 2026 with a full Photo page, nine AI tools, and a free download that directly challenges Adobe subscriptions.
Cursor 3.1 Adds Tiled Multi-Agent Layout and Voice Input
Cursor shipped version 3.1 on April 13, adding tiled multi-agent layouts, upgraded voice dictation, and a string of performance fixes that make the AI coding editor feel noticeably faster.
Seedance 2.0 Arrives in ComfyUI: What It Means for Video Creators
ByteDance's Seedance 2.0 is now a native node in ComfyUI, bringing multimodal audio-video generation with 12 simultaneous reference inputs directly into the most popular open-source creative AI pipeline.
Springboards Flint Scores 7/10 on AI Novelty Bench
Springboards launched Flint, a creative AI model engineered for novelty that generates seven functionally distinct responses per ten prompts, compared to 2.88 for leading LLMs.
Meta Builds Zuckerberg AI Clone, Plans Creator Digital Twins
Meta develops photorealistic AI avatar of Zuckerberg for 72K employees. Superintelligence Labs tech could let creators and influencers build their own digital twins.
Stanford AI Index 2026: What 53% Adoption Means for Creators
Stanford HAI's ninth annual AI Index Report reveals that generative AI reached 53% global adoption in three years, while the creative AI sector faces a transparency crisis and environmental reckoning.
ComfyUI v0.19.0 Adds Music, Text Gen, and Video Nodes
ComfyUI v0.19.0 lands with Ace Step 1.5 XL music generation, Qwen 3.5 text generation, SeeDance 2.0 video nodes, and a faster Flux 2 decoder.
llama.cpp Adds Audio Support With Qwen3-Omni
llama.cpp release b8769 adds audio multimodal support for Qwen3-Omni and Qwen3-ASR models, bringing local speech recognition and audio understanding to consumer hardware.
VoxCPM2 Open-Sources 2B-Param TTS With 30 Languages
OpenBMB released VoxCPM2, a 2 billion parameter text-to-speech model that runs on 8GB VRAM, supports 30 languages at 48kHz, and can design voices from natural language descriptions.
Claude for Word Beta Adds AI Editing to Microsoft Word
Anthropic launched Claude for Word in public beta, bringing AI-powered document editing with native tracked changes to Microsoft Word on Mac and Windows.
MiniMax Music 2.6 Adds AI Cover and Style Transfer
MiniMax released Music 2.6 on April 10, adding a Cover feature that extracts a song's melodic skeleton and lets creators rebuild everything around it.
Game Studios Hit 4-20x Faster With AI
Game studios reorganizing around AI are completing prototypes 4x faster and generating UI assets up to 20x faster, according to new Wharton research.