What auto-editor does: cut silence automatically and export to Premiere or Resolve

Thu, 23 Apr 2026 13:31:27 +0800

What takes the most time in many videos is often not transitions, color work, or subtitles. It is watching a long source file from start to finish and manually removing all the parts where nobody is speaking, nothing is moving, and nothing useful is happening.

That is the problem auto-editor is built to solve. It is not a traditional visual editing app. It is a command-line tool that analyzes video or audio first, then automatically cuts silent sections and low-activity segments so you can finish the first rough pass much faster.

This kind of tool is especially useful when your source material looks like this:

screen recordings
tutorial videos
podcasts or talking-head recordings
long livestream replays
interview footage that needs a rough pass before fine editing

What is its core value

The main idea behind auto-editor can be reduced to one sentence:

automate the boring first pass, then hand the result back to a human.

The official README emphasizes that it removes common “dead space”, especially silence. For long videos, this is often the most repetitive and mechanical part of the whole editing process.

The most basic usage is very simple:

`1`	`auto-editor path/to/your/video.mp4`

After it runs, you can think of the result as a first round of filtering. It is not making the final creative decisions for you. It is clearing out the parts that obviously contain little useful information.

It does not only cut by silence

When people hear “automatic editing”, many assume it only detects silence.

But auto-editor supports more than one editing method. In the official docs, the two common approaches are:

audio: judge sections by loudness or volume
motion: judge sections by visual motion

For example:

`1`	`auto-editor example.mp4 --edit motion:threshold=0.02`

That matters because the tool is not limited to footage where the only rule is “cut when no one is talking”. It can also work for videos where the important signal is visual activity rather than speech alone.

The official examples also show that you can combine methods, such as using both audio and motion conditions together. That makes it easier to express something closer to a real editing rule instead of relying on a single threshold.

`--margin` matters more than it first appears

The easiest way for an automatic rough cut to feel wrong is not failing to remove silence. It is cutting too aggressively.

If the first word of a sentence is clipped, or the end of each line is cut off too sharply, the result starts to feel mechanical. auto-editor has a practical option for this: --margin.

You can think of it as adding a small buffer around the sections that would otherwise be trimmed away. In the official examples, the default value is 0.2s, which means a bit of padding is left around the cuts so the pacing feels more natural.

For example:

1
2

auto-editor example.mp4 --margin 0.2sec
auto-editor example.mp4 --margin 0.3s,1.5sec

This option is worth paying attention to because it often determines whether the output feels like a harsh automatic cut or a decent human-style first pass.

It is best understood as a first-pass tool, not a full NLE replacement

This is the most important point when judging auto-editor.

It behaves more like an automated rough-cutter than a complete non-linear editor. Its strengths are:

processing long footage quickly
automating repetitive dead-space removal
compressing obviously low-value stretches before real editing starts

But once you need these things, you will probably still return to a traditional editor:

fine rhythm adjustments
manually picking the best moments
subtitles, transitions, and B-roll
color, audio polish, and final packaging

So the realistic framing is not “you no longer need Premiere or Resolve”. It is “let auto-editor do the repetitive labor first, then move into a real editor for finishing work”.

It can render directly or export timelines to other editors

This is another reason the tool is practical.

If you only want a quickly cleaned-up version with dead space removed, it can render media directly. But if you plan to continue editing, the official docs also support exporting timelines to common editors.

The README lists these export targets:

Adobe Premiere Pro: --export premiere
DaVinci Resolve: --export resolve
Final Cut Pro: --export final-cut-pro
Shotcut: --export shotcut
Kdenlive: --export kdenlive
individual media clips: --export clip-sequence

For example:

`1`	`auto-editor example.mp4 --export premiere`

That is useful because it makes auto-editor feel like a front-end processing tool rather than a closed workflow you are forced to stay inside.

It also supports manual rules

Many automation tools become frustrating as soon as they make a wrong decision.

auto-editor is more flexible than that. It lets you explicitly mark certain sections to always cut or always keep. The common examples in the official docs are:

--cut-out: always remove a section
--add-in: always keep a section

For example:

1
2

auto-editor example.mp4 --cut-out 0,30sec
auto-editor example.mp4 --add-in 0,30sec

That means you do not have to give all control to the program. You can let automatic rules handle the bulk of the work and then use manual rules to correct edge cases.

The official install recommendation is now the binary release

According to the official install page, the recommended path today is not pip, but the prebuilt binaries from GitHub Releases.

The rough process is:

Download the binary for your platform from the Releases page.
Rename it to auto-editor or auto-editor.exe on Windows.
Run it from your terminal.
Use auto-editor --help to verify the installation.

The official page also mentions package-manager installs:

on macOS: brew install auto-editor
on Ubuntu or Debian-like systems: apt
on Arch-like systems: yay -S auto-editor

One detail worth noting is that the install page explicitly says new versions are no longer being published to pip, so pip install auto-editor still exists, but it is no longer the preferred path.

If `yt-dlp` is installed, it can take URLs as input

There is also a useful optional dependency note on the install page: if yt-dlp is installed, auto-editor can use URLs as inputs directly.

That is convenient for workflows where you first fetch online footage and then run an automatic rough cut. In other words, it does not only work with local files.

The more practical takeaway, though, is still to make sure your source and usage rights are clear before folding those materials into your editing workflow.

Who is this tool best for

If your work depends heavily on storytelling rhythm, shot language, and lots of human editorial judgment, auto-editor will not replace the hardest part of the craft.

But if your work often looks like this, it can save real time:

removing long pauses from an hour-long screen recording
stripping silence from a spoken recording before moving into an editor
making a first rough cut of course content before adding subtitles
batch-processing long materials with similar structure

Its value is not that it edits with better taste than a person. Its value is that it automates the most repetitive and patience-draining first step.

Conclusion

What makes auto-editor worth paying attention to is not that it turns editing into some magical AI black box. It is that it solves one narrow problem honestly: how to finish the first rough cut faster.

If you treat it as a front-end tool for removing silence, trimming dead space, and exporting timelines, its role becomes very clear:

it is not responsible for final editorial taste
it is well suited to the first pass on long footage
it complements editors like Premiere and Resolve instead of replacing them

For many tutorial, screen-recording, talking-head, and long-form video workflows, that kind of tool is not flashy. It is simply the kind that saves time on repetitive labor.

Productivity Tools on KnightLi Blog

What auto-editor does: cut silence automatically and export to Premiere or Resolve

What is its core value

It does not only cut by silence

--margin matters more than it first appears

It is best understood as a first-pass tool, not a full NLE replacement

It can render directly or export timelines to other editors

It also supports manual rules

The official install recommendation is now the binary release

If yt-dlp is installed, it can take URLs as input

Who is this tool best for

Conclusion

`--margin` matters more than it first appears

If `yt-dlp` is installed, it can take URLs as input