Speed-up your videos speeding-up (or removing) silences, using FFmpeg.
Read this in other languages: English, Italian.
This program, built with Electron, makes use of FFmpeg in order to speed up (or skip) parts of the video that are with no sound.
It is really helpful for video lessons, where the professor takes long time for writing something on the board, makes lots of pauses, or just speaks slowly. With this program you can skip this boring parts and save time.
Of course this program is not perfect, and you might have to practice a bit with it.
Import your videos, choose the program settings and press Start. The app shows you the progress status, which consists of 3 steps: Detecting silences, Exporting and Concatenating.
Not all videos are with the same audio volume, and you may want to choose which silences should be treaten as that. So, here we have the configurable parts.
These settings change the way FFmpeg detect silences. You can set:
-
The background noise of the video (
Lowfor a silent room with a microphone,Midfor the average noisy room,Highfor a noisy room).Note: if your are new, try a video with defaults settings and see the result.
-
How many seconds the smallest silence lasts (this value prevent brief pauses to be treaten as silences).
-
How many seconds of silences should not be treaten as silences.
Explaination: without a minimum time of margin, spoken words might merge and the result would be an incomprehensible speech.
With these settings you can change the speed of spoken/silence parts of the video, and also set silence parts to be video-only.
Choose the format (extension) of your video (default is set to keep the same extension as the input file). Other settings include fps (frame per second), cfr and preset (these are settings for the video codec h264). The codec used for audio is aac.
Note: the default path of the videos (as well as temporary files) is set to <your home path>/speededup/. If you want to change it, press the settings button or go to File -> Settings.
This program does not need to be installed to run, as I packaged it with electron-packager. But, if you want to compile and run this program by yourself from the source code, then:
$ git clone https://github.com/padvincenzo/silence-speedup
$ cd silence-speedup
$ npm install
$ npm start
If you also want to embed the FFmpeg binaries, here is explained how to.
Note: if you wish to run this program from source files then you also need to install NodeJS.
For each video, this program will:
- Run ffmpeg with
silencedetectfilter, in order to get the list of silences' start/end timestamps.
<ffmpeg bin> -hide_banner -vn \
-ss 0.00 -i <Input file> \
-af silencedetect=n=<threshold>:d=<duration> \
-f null -
- Using that list, split the original video in a tmp folder, applying a speed filter, if any.
<ffmpeg bin> -hide_banner -loglevel warning -stats \
-ss <Start time> -to <End time> -i <Input file> \
-filter_complex "[0:v]<setpts filter>[v];[0:a]<atempo filter>[a]" \
-map [v] -map [a] <Output fragment>
- Concatenate all the fragments generated before.
<ffmpeg bin> -hide_banner -loglevel warning -stats \
-f concat -safe 0 \
-i <Fragment list file> \
-c copy \
-map v -map a <Output file> -y
At the end of execution, the program does not automatically clean the tmp folder.
Anyone can contribute to this project, in many ways:
- Sharing the project;
- Translating this project in other languages;
- Finding and report/fix bugs;
- Suggesting new ideas;
- Implementing new functionalities.
For any doubt or perplexity we can discuss here.
This software uses binaries of the FFmpeg project, which I do not own, under the GPLv3.