Best AI applications: Tools that you can run on Windows, macOS, or Linux

If you have a powerful PC with a good GPU, then you've probably been enticed by the world of AI. There are so many tools out there that you can run on any computer with Windows, macOS, or Linux, like Stable Diffusion, LM Studio. and more. These are some of our favorites that anyone can download and install today for free.

LM Studio

Converse with any LLM

Large language models (LLM) like ChatGPT, Google Bard, and Microsoft Copilot all run in the cloud, and that basically means they run on somebody else's computer. Not only that, they're particularly costly to run, and that's why companies like OpenAI and Microsoft are bringing in paid subscription tiers. However, you can run many different language models like Llama 2 locally, and with the power of LM Studio, you can run pretty much any LLM locally with ease.

If you want to run LM Studio on your computer, you'll need to meet the following hardware requirements:

• Apple Silicon Mac (M1/M2/M3) with macOS 13.6 or newer
• Windows / Linux PC with a processor that supports AVX2 (typically newer PCs)
• 16GB+ of RAM is recommended. For PCs, 6GB+ of VRAM is recommended
• NVIDIA/AMD GPUs supported
• An (optionally fast) internet connection to download models

Once you meet those requirements, you're good to go! LM Studio can converse with major LLMs like Llama 2, Mixtral 8x7B, Zephyr, and more.

Stable Diffusion

Generate images

If you want to generate images for the likes of a Dungeons and Dragons campaign, wallpapers for your phone and PC, or anything else really, then Stable Diffusion is the tool to do it. While shrouded in controversy, it's a powerful tool for those who want to use it for their own personal projects. With a local set up, there are forks such as AUTOMATIC1111's Stable Diffusion WebUI that come with a built-in upscale tool that can increase the resolution up to four times higher. While you can generate images at higher resolutions, it is often much quicker to generate an image at a lower resolution and then upscale it. All of the images below are upscaled from smaller resolutions.

All you need is a lot of vRAM and you're good to go, as these models require a lot of it. I can generate images quite quickly on an RTX 4080, but you'll be able to on weaker cards, too.

Final2x

For upscaling images quickly

If you have older images that are a lower resolution, Final2x might help you increase the quality of those images. A simple tool that you can drag and drop images on to, you'll be able to upscale images to twice their resolution. It runs on Windows, macOS, and Linux, and we tested it out to see how well it did. This is the original image:

I then halved the size of the image and ran it through Final2x to upscale the image back to the original resolution.

Not bad at all, right? This is a free tool that anyone can use, and it's a great way to upscale images quickly before sending them off to a friend. On my M1 Pro MacBook, it took less than ten seconds to upscale that image.

Whisper

Transcribe anything quickly and accurately

Several methods exist for transcribing interviews or videos. One approach involves manual transcription through listening, which, while time-consuming, ensures the highest level of accuracy. Alternatively, automated services or tools can be utilized. Previously, I used YouTube's auto-captioning feature, downloading the generated subtitles and refining them to correct any inaccuracies. At the moment, numerous AI-powered solutions are available that perform exceptionally well, including OpenAI's Whisper.

Whisper from OpenAI is an open-source tool that you can run locally pretty easily. It works on Mac too, though there are some more convoluted steps to get it working. You'll basically just need to compile a C++ version of Whisper from the source yourself. It's not an official port, but it's the only way to get it to run natively on Apple silicon. You can follow this tutorial on Medium for how to do that.

if you ever need to do some transcription work, I highly recommend Whisper. It's helped me in countless situations already, and especailly for content creators looking to generate captions, it's a gamechanger.

Spleeter

Get instrumentals from any song

If you're as big of a music fan as I am, sometimes you may just want to listen to the instrumental of one of your favorite songs. That's why Spleeter is a great tool, as it allows you to split a track into multiple "stems." This tool has actually been around for years, and it can separate a track into up to five separate stems: vocal, piano, drums, bass, and "other." If you just want an instrumental separated from vocals, there's a two-stem model that does that, too.

In my testing, I found that more simplistic tracks like Frank Sinatra's Fly Me to the Moon will come through perfectly, whereas other songs that are more hectic, such as Modern Baseball's The Thrash Particle will come through pretty well, especially in its instrumental, but the vocal component may sound empty (albeit fine enough).

Spleeter is a tool developed by Deezer, and you install it by running the following commands.

pip3 install ffmpeg-python

Once done, you can install spleeter by running the following command:

pip3 install spleeter==2.0

This will add Spleeter to your Python "Scripts" folder, so you can either add that to path, or use the following command prefix every time you want to use Spleeter:

python3 -m spleeter 

There's something for everyone

If you're looking to embark on an AI journey, then these are some of the best tools you can start off with. There really is something for everyone, be it music lovers, art lovers, or even just people who are interested in technology.