Increase or normalize the volume of a video
A video shot from far away or a screen recording often has audio that is too quiet. You can multiply the volume by a fixed factor, or automatically normalize it to a standard level. In both cases, the picture is copied without modification.
Updated on
How to use it
- Install ffmpeg if you haven't already.
- Open a terminal in the folder containing your file.
- Copy the generated command, adjust the file names if needed, then run it.
- Open the resulting file to check the result.
How to measure the volume of a video
The volumedetect filter displays the mean volume and the maximum peak, without creating a file. A peak close to 0 dB means you can't increase the volume any further without distortion.
ffmpeg -i "video.mp4" -af volumedetect -vn -f null -
Single-pass normalization
The command uses loudnorm in a single pass, which is enough in most cases. For maximum accuracy (podcasts, broadcasting), loudnorm can be used in two passes: a first one to measure, then a second one that applies the measured values.
Frequently asked questions
What is the difference between increasing and normalizing?
Increasing multiplies all the audio by the same factor: passages that are already loud can distort. Normalization (the loudnorm filter) analyzes the perceived loudness and adjusts it toward a standard target level (-16 LUFS, used for the web and podcasts), while avoiding distortion.
The audio crackles after the increase. Why?
The factor is too high: the peaks exceed the maximum and get clipped. Choose a lower factor or automatic normalization.
Is the picture re-encoded?
No: the "-c:v copy" option copies the video as is. Only the audio is processed, which makes the operation fast and lossless for the picture.