Morrolinux: AI, the FOSS Way

Morrolinux: AI, the FOSS Way

Harnessing the Power of Open Source AI: Free Tools for Creativity and Productivity

Whether you’re editing images, isolating vocals, transcribing audio, or experimenting with Large Language Models (LLMs), open source AI offers some of the best solutions available today. From tools designed for creative tasks to those that boost productivity, here’s a breakdown of my favorite AI-powered software and why they matter.

Image Manipulation: Upscayl, rembg, and Fooocus

First up is Upscayl, an open source gem for upscaling low-resolution images. Need to sharpen an image downloaded from the web? Upscayl runs locally and delivers impressive results. I use it often to enhance visuals for my thumbnails, achieving higher quality in just seconds. The tool requires a GPU.

For background removal, rembg is my go-to. It’s simple: drag an image, click, and voilà: background removed! You can even automate the process with command-line tools, making batch editing a breeze.

Looking for creative flair? Check out Fooocus, an image generation tool similar to MidJourney. It allows you to expand backgrounds, remove objects, and more. Whether you’re creating a holiday scene or making edits for fun, Fooocus is a powerful and versatile solution.

Audio Tools: Ultimate Vocal Remover and Whisper

Need to separate vocals from background tracks? Ultimate Vocal Remover (UVR) has you covered. As a content creator, it has saved me countless hours by isolating clean audio from noisy recordings.

For transcriptions, I rely on Whisper by OpenAI. It’s an open source model that converts speech to text with surprising accuracy. Whether it’s subtitling videos or transcribing interviews, Whisper integrates seamlessly into workflows with its API (multilanguage) support.

Exploring Large Language Models (LLMs)

LLMs like ChatGPT have transformed how we approach tasks such as content generation, text summarization, and coding assistance. But did you know there are open source alternatives that rival ChatGPT?

  • LLAMA3 by Meta: Available in configurations from 8 billion to 70 billion parameters, this model is lightweight enough to run locally on consumer GPUs.
  • Mixtral 8x22b: A standout for coding tasks, Mixtral impressed me with its ability to generate Python scripts, handle dynamic inputs, and store data persistently.

Small, specialized models like Qwen2.5 7b excel in specific domains, offering efficiency and reliability without the hardware demands of larger models.

Integrating AI Models into Everyday Tools

AI doesn’t stop at standalone apps. Integrations are where the magic happens. With plugins like Continue for Visual Studio Code, you can pair LLMs with your development environment to obtain code suggestions, explanations, and refactoring.

Browser-based assistants like Brave’s Leo add AI-powered features directly into your browsing experience. And for command-line enthusiasts, ShellGPT brings AI into your terminal, streamlining tasks such as file searches, system updates, and even generating Git commit messages.

Open Source Is the Future

Open source AI tools not only democratize access to cutting-edge technology but also empower creators and developers to innovate on their terms. These tools are robust, flexible, and often just as powerful as their proprietary counterparts. They also offer a sense of security: knowing the code is open for inspection means there are no hidden surprises or risks to your privacy.
Let me know in the comments which of these tools you’ve tried or whether there’s an open source gem I missed. I’m always on the lookout for new AI tools to test and review. See you in the next Morrolinux Tips!

Watch in Italian the Full Demo and a complete walkthrough of these tools: Ho provato i migliori software AI Open Source (I tested the best open source AI software). Stay tuned for more tips and tutorials in the Morrolinux series!

<< Read the previous post of this series

About Moreno Razzoli:

My name is Moreno Razzoli. I have a degree in Computer Science, and I hold certifications in Linux from LPI, CompTIA Linux+, and Suse CLA. I am also an authorized Training Partner of the Linux Professional Institute. I have worked on various Open Source projects and have contributed to several existing projects on GitHub. Since 2008, I have been creating educational videos on YouTube and my official website.

發佈留言

發佈留言必須填寫的電子郵件地址不會公開。 必填欄位標示為 *