Interested in artificial intelligence? It is a field that moves faster than humanly possible. To help keep you up to speed, here are some of the latest AI news updates.

ElevenLabs Expands from Voice Synthesis to Sound Effects

After OpenAI broke the internet showing off their phenomenal Sora text-to-video capabilites, ElevenLabs followed up with text-to-sound effects, cheekily showing how the Sora publicity videos would be well-complemented by some immersive sound effects.

If you watch a lot of movies or TV, you’ll often hear the exact same sound effects, sometimes from 30-year-old libraries, used over and over again. Now, you’ll be able to go down unexplored foley paths, just as we’ve come to expect with text and code, images, and video.

Having successfully cloned my and other voices with ElevenLabs, I can’t wait to try generating sound effects. But, as with Sora, there’s a waiting list!

Microsoft Invests €3.2 Billion on AI in Germany

Microsoft is investing 3.2 billion euros to significantly expand AI infrastructure and cloud capacities in Germany, marking its largest investment in the country in 40 years. This investment aims to more than double the current capabilities, particularly through expansions in Frankfurt and new infrastructures in North Rhine-Westphalia. By the end of 2025, Microsoft also plans to train over 1.2 million people in digital skills, enhancing Germany’s position in AI technology and supporting the structural transformation in various economic sectors. The investment is seen as a boost for the German economy, leveraging AI to maintain its global competitive edge and meeting the rising demand for AI applications across different industries.
 


Microsoft’s initiative includes a focus on responsible AI and sustainability, aiming for 100% renewable energy use by 2025. This move has received positive reactions from German officials hoping to advance the country’s digital infrastructure and AI ecosystem. For more details on this transformative investment and its implications for Germany's AI future, you can read Microsoft’s (German) press release here.

Subscribe
Tag alert: Subscribe to the tag Embedded & AI and you will receive an e-mail as soon as a new item about it is published on our website!

NVIDIA Chat With RTX: Easier Local Chatbots

NVIDIA CEO Jensen Huang advocated for national AI sovereignty at the World Government Summit in Dubai as NVIDIA’s market value has surged to an impressive $1.8 trillion USD, overtaking Amazon in the process.

Chat with RTX is their new tool — currently available only for Windows users — designed for local, offline use, allowing you to interact with various documents on your computer without the need for an internet connection.
 
“Chat With RTX,” NVIDIA’s new local GPT tool. Source: NVIDIA.
This ensures that your data remains private and isn’t uploaded anywhere. It supports open-source models such as LLaMA and Mistral, featuring a straightforward and user-friendly interface. However, it’s worth noting that it requires a fairly powerful computer to run effectively, as it leverages advanced technologies such as TensorRT-LLM and Tensor Cores for improved speed and efficiency compared to other offline AI tools.
 


This tool offers the ability to chat with your documents in formats like .txt, .pdf, .docx, and .xml, and even interact with YouTube video transcripts (though fetching transcripts requires an internet connection). It's a 100% private way to utilize AI, potentially justifying an upgrade if your current setup isn’t up to the task. You can download and try it out for yourself at NVIDIA’s website.

Subscribe
Tag alert: Subscribe to the tag Nvidia and you will receive an e-mail as soon as a new item about it is published on our website!

French AI Startup Announces mistral-next Large Language Model

Paris-based Mistral AI was founded in April 2023. It advocates for open model access, allowing developers complete control over their applications’ engines. By the end of 2023, it had raised $415 million and was valued at over $2 billion. The company focuses on generative AI models.

Recently, it launched mistral-next, a language model designed to compete with GPT-4, with a focus on advanced features and ethical AI. It is said to stand out for its exceptional text generation, producing contextually relevant and coherent responses. It should show superior performance, especially in tasks requiring deep understanding and creativity. Testers highlight its capability to handle complex problems, suggesting it could rival GPT-4, but we’ll find out for ourselves when it is released widely, which should happen very soon!