OpusClip’s Post

Explore ClipAnything here: https://lnkd.in/gq5K4x6Z

Automatic clipping is not a unique feature, most repurposing tools nowadays have had it. However over the last year I have used Opus, and the amount of usable clips has only gotten increased, and higher quality. Especially compared to other tools on the market. Now that Opus is going to be able to "Clip Anything" as its called. This means more than just talking videos like podcasts or talking head content. From sports, to news, and even gaming after some refinement. I think this is a really cool update, and wanted to share! #opusclip #contentrepurposing #polytools

I hear a lot these days about AI empowered tools to detect emotion in spoken language. I must say that the examples I've seen so far are not, well, inspiring. It looks (or it sounds? 🙄 ) that they still have to rely more on the words being spoken than on the actual tone and inflection. I played dirty and I said something negative with a positive tone, and viceversa, and the system still will catch the content more than the style. But content (suprise, surprise) is often not enough. "Sentiment analysis" has gained quite a negative reputation for its ambition to catalogue positive and negative and anything else, as words alone are not enough to detect attitude and therefore meaning. "Your product rocks" can be said with a voice that implies enthusiasm; sarcasm; anger; and anything in between. Now, if we could really combine the two (what you say, and how you say it) to read into the speaker's mind, that would rock. The potential applications are countless, from call centers to politics. Shall we get there? Are we already there? Am I missing something? Feedback appreciated.

Interesting article on how AI can't match human voice, humor, reasoning, and nuance. The quote below makes an interesting comparison to vinyl versus CD. It's well thought out but I have to say that, even within the 2 years I've been playing with Generative AI, I've seen it become more and more lifelike. Today, I found myself feeling a bit guilty for not thanking ChatGPT after it made hundreds of scenarios based on made-up Account numbers I fed it: "When CDs first came out, they were touted as a perfect digital copy of music. The digital version, however, was an interpretation of music by engineers: Bits of information were left out that captured the unique timbre of a vocal or drumbeat; complicated frequencies were abridged into simplified code. Decades later, we all recognize that vinyl—sound waves pressed in plastic, scratched into music through a needle—is closer to the original recording. In the same way, GPT compression—and other summarization tech—robs us of voice, humor, reasoning, the soft nuance that comes with actual human thought."

OpenAI is pushing boundaries once again with GPT-4o, and it's truly mind-blowing to watch two AIs sing a duet in real time 🤯 They might need a bit more practice before auditioning for The X Factor, but this sh** is incredible. The rapid progress in AI creativity raises intriguing questions about the future of human ingenuity in music. Can we get these questions answered while this tech is advancing (not likely)?! 🤷🏾♂️ #AI #MusicTech #ArtificialIntelligence #Radio

Every week, I set out time to dive into 4 to me close topics that are shaping my life and our future: Climate, Web3, Design, and AI. This week’s AI exploration brought some pretty exciting discoveries—Mistral’s new update and Luma’s latest release. If you’ve been following along, you know I’ve already talked about ChatGPT and Claude AI , the big names in text-based LLMs. But let’s talk about Mistral AI - this Europe-based solution just got a major upgrade, and it's looking like a serious contender. Mistral AI's new ‘Agents’ feature - You don’t need to be a coder to build your own AI model anymore. Here’s how you can get started: 1. Sign up on Mistral AI's website and hit "Build on La Plateforme." It’s about as easy as setting up a Netflix account. 2. Click on “Create an Agent,” and choose your AI model—Mistral Large 2 is my pick because, well, why not go big? 3. Adjust the randomness slider to control how creative your AI gets. Think of it like tweaking your playlist on Spotify. 4. Add example interactions if you want your AI to really nail the responses (totally optional, but worth it). 5. Deploy - and start testing. It’s that simple. And then there’s Luma AI Dream Machine 1.5”—this one’s got my inner designer pretty excited. As someone who’s always been passionate about design and creativity, this tool is next level for AI video generation: - Dream Machine 1.5 can create 5-second video clips from text and images that look pretty damn realistic. - It’s got improved natural language processing, which means your prompts are understood and executed more accurately. - This tool is all about smooth motion and cinematic flair—perfect for turning static ideas into dynamic stories. - But in reality, it’s still got some catching up to do on things like morphing and movement. But we’re getting there. Imagine creating Hollywood-level clips from just a few lines of text. We’re not quite there yet, but with Luma Labs pushing the envelope, it’s only a matter of time. So, what’s your take? Ready to dive into AI, or are you still on the sidelines? Let’s chat.

A lot of buzz about agents and Agentic AI recently. This is a great overview/intro to what they are and their potential applications by Tobias Zwingmann and Dr. Tristan Behrens. I read this on Tobias' newsletter, The Augmented Advantage. Highly recommend subscribing to it - lot's of great content to explore.

🌐 Discover the latest in AI! Watch my new video showcasing a Huggingface Spaces app that ranks LLMs on a dynamic leaderboard. 📈 🚀 Dive into the competitive world of language models and see how they stack up against each other! hashtag #AI hashtag #TechInnovation

🌐 Discover the latest in AI! Watch my new video showcasing a Huggingface Spaces app that ranks LLMs on a dynamic leaderboard. 📈 🚀 Dive into the competitive world of language models and see how they stack up against each other! hashtag #AI hashtag #TechInnovation

𝗪𝗵𝗮𝘁 𝗱𝗼 𝗜 𝘁𝗵𝗶𝗻𝗸 𝗮𝗯𝗼𝘂𝘁 𝗥𝗔𝗚? (𝗥𝗲𝘁𝗿𝗶𝗲𝘃𝗮𝗹 𝗔𝘂𝗴𝗺𝗲𝗻𝘁𝗲𝗱 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝗼𝗻) They need more marketing. RAGs can solve some problem scenarios except the one we currently have. Right now, we have billion-dollar marketing and PR efforts from major LLMs saying, "These are the most capable AI tools on the planet." Hundreds of millions of unknowing users don't know what they don't know. They hear big tech companies proclaiming, "We are brilliant, and our tools are the most capable on the planet." In the search industry, Google made this same play. Google claimed to be everything to everyone. Even when better niche search solutions existed, everyone just "Googled it" and failed to educate themselves about more effective solutions. The real challenge lies in educating most users to understand that the solution isn't always the big, sparkly thing being touted on stage. When we start seeing billion-dollar marketing efforts from RAG, multi-agent, or other advanced AI systems we must emphasize the importance of critical thinking and due diligence in technology adoption. Instead of blindly following the loudest voices, users should explore diverse options, understand their unique needs, and select tools that genuinely enhance their productivity.