tsplice

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

★ 25 Open GitHub ↗

Splice and merge videos from the terminal :scissors: :film_strip:

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# tsplice

A modest command-line application to **splice** and merge video files from the **t**erminal.

![](https://raw.githubusercontent.com/aschmelyun/tsplice/main/art/screenshot.png)

:construction: This is an experimental app so you might see some weird bugs. If so, [let me know](https://github.com/aschmelyun/tsplice/issues)! :construction:

## Installation

To install, download the latest [release binary](https://github.com/aschmelyun/tsplice/releases/latest) to your system. If you're on Linux or MacOS, you can use this one-liner to handle that:

```
sudo curl -L https://github.com/aschmelyun/tsplice/releases/download/v1.0.3/tsplice-v1.0.3-darwin-arm64 -o /usr/local/bin/tsplice && sudo chmod +x /usr/local/bin/tsplice
```

> [!NOTE]
> The release above is for Apple Chip MacOS machines, be sure to change the link in the curl call to the appropriate binary for your system.

## Requirements

You'll need to have the following software installed on your system to use `tsplice` effectively:

- ffmpeg
- mpv

Additionally, you'll need to have an [OpenAI API key](https://platform.openai.com/api-keys) ready to be set on the first run.

## Usage

Run `tsplice` in any terminal window, followed by the file that you want to edit.

```sh
tsplice ./Movies/my_facecam_vid_20250629.mp4
```

Additionally, you can pass in some options between the command and the video file. The ones available are:

- `lang`: (optional, string) sets the language for the transcription, if you want to prompt the model to make it easier to transcribe
- `prompt`: (optional, string) sets a prompt for the Whisper model, if you want to provide extra context to the model during transcription
- `gate`: (optional, bool) removes blocks of 10+ seconds of silence from the audio before sending off for transcription

A command using some of these might look like:

```sh
tsplice --gate --lang=fr ./Movies/my_facecam_vid_20250629.mp4
```

After running through the initial steps of extracting audio and transcribing with Whisper, you'll be presented with a list of lines from your video's audio that you can toggle to select or deselect.

You can press `p` at any time to see a pop-up preview of that current line using your original video.

Press `c` after you've selected all of the clips that you want in your final video to start the merge process.

At any time, you can get a help screen by running just `tsplice` or `tsplice --help`. You can see the current version installed by running `tsplice --version`.

## How it works

This app performs a few basic steps:

1. Extract audio from the video using `ffmpeg`
2. Send the audio to OpenAI's Whisper API for transcription
3. Save the transcription to a local file
4. Parse the transcription into individual lines and add it to a checklist
5. Take the selected checklist items and compile them to a list of timestamps
6. Merge together the final video with `ffmpeg` and the timestamp list above

That's it! Your spliced video will be available in the same directory as your original.

tsplice

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README