NoiseTorch

One of the most common problems we all encounter with remote video calls and conversations is background noise. Even the quietest home office is easily plagued by the sound of fans, air conditioners, children, brewing coffee, and distant traffic. Some chat platforms, such as Google, do offer some kind of noise reduction, but they never give you any control over its parameters. NoiseTorch is a very clever solution to this problem that both reduces noise and gives you complete control over the process, Remarkably, it does this within the confines of PulseAudio. It's an audio processor that sits between your microphone, or audio input, and the application expecting the audio input. This is why, when NoiseTorch is installed, you're first asked to select an input device to monitor. The UI even helpfully filters out the various display monitor sources that litter so many of our inputs, after which you can launch NoiseTorch proper.

The great thing about using PulseAudio is that NoiseTorch works with any application, because as soon as it's running, PulseAudio thinks it has another input source. This appears as NoiseTorch microphone in your applications, but it's really just a virtual input using your previously configured input filtered through NoiseTorch's noise reduction algorithm. Your original input or microphone is also selectable, which is great if you need to quickly revert to an input without the noise reduction. The algorithm NoiseTorch uses is called RNNoise, a product of deep learning that's hosted by Mozilla Research. It's great to see a static open source library being used for something practical, especially when a closed source input filter could so easily be subverted. Other than input selection, there's only a single value to adjust while a call is in progress. This acts as a noise gate, cutting the sound completely when it falls beneath a certain level. You won't directly appreciate the quality of the end result, but your colleagues and family on the other end will.

Project Website: https://github.com/lawl/NoiseTorch

NoiseTorch is a brilliant way of reducing the level of background noise from your microphone (or any other input), without further dependencies.

Markdown editor

Aurora Editor

We've looked at plenty of minimal Markdown editors in these pages, but the shiny new Aurora Editor is showing plenty of promise. It even lowers expectations by describing itself as "yet another lightweight Markdown editor," which still never fails to win us over. First impressions indicate that there's not that much difference between Aurora and the average editor, however, with the UI being even more minimal than most. The main view is split horizontally between a left pane and a right pane, and the left pane is where you do your work and actually edit the text. This pane has a simple geometric texture on its background that can't be changed, but it does put you in a scientific frame of mind while writing. You can select between a dark and a light theme from the View menu and toggle full-screen mode. These are the only configurable options, but it doesn't mean Aurora isn't comprehensive in its editing ability.

The right pane shows a preview of the HTML rendered output. This is easy for text, but what makes Aurora unique is its ability to render more than the simple HTML prerequisites of lists, emoji, and tables. You can include mathematical formulas written in MathJax, which are rendered perfectly in the preview. Its source code highlighting is some of the best we've seen, including the difficult-to-parse territory of JSON. The latter even includes folding, so you can hide a specific hierarchy of elements in the preview. But you can even include Graphviz and Mermaid source directly in your Markdown, and they'll be rendered perfectly in the preview. Most importantly, if you choose to export your Markdown as a single HTML file, all this beautiful output is maintained, which means you can easily use Aurora to produce user-friendly output and documentation for websites. As this is such a functional early release, we can't wait to see what developments follow.

Project Website

https://aguang-xyz.github.io/aurora-editor

Regardless of which elements you embed within your Markdown, Aurora Editor seems to create a perfectly rendered preview.