Modulate FFmpeg Volume Based on Time

This article explains how to dynamically modulate audio volume over time using FFmpeg’s volume audio filter. You will learn the correct syntax to enable continuous evaluation, the key variables available for your mathematical formulas, and practical command-line examples for creating fades and oscillating tremolo effects.

Enabling Time-Based Evaluation

By default, FFmpeg evaluates the volume filter’s expression only once at the beginning of the stream. To change the volume dynamically as the file plays, you must set the eval parameter to frame. This forces FFmpeg to re-evaluate your equation for every audio frame.

The basic filter syntax is:

-af "volume='EXPRESSION':eval=frame"

Key Variables for Equations

Within your expression, you can use several built-in FFmpeg variables: * t: The current timestamp of the audio frame in seconds. * n: The sequential number of the audio frame. * pos: The position of the frame in the input file (in bytes). * PI: The mathematical constant \(\pi\) (approx. 3.14159).

Practical Examples

1. Simple Tremolo (Sinusoidal Modulation)

To create an undulating volume effect (tremolo) that oscillates between 20% and 100% volume at a frequency of 2 Hz (twice per second), use the sine function:

ffmpeg -i input.mp3 -af "volume='0.6+0.4*sin(2*PI*2*t)':eval=frame" output.mp3

• 0.6 is the baseline volume offset.
• 0.4 is the amplitude of the oscillation (modulating the volume by \(\pm 0.4\)).
• 2*PI*2*t defines the frequency of 2 Hz.

2. Custom Fade-In

While FFmpeg has a dedicated afade filter, you can achieve a custom mathematical fade-in using the volume filter. To linearly increase the volume from 0 to 1 over the first 5 seconds of a file:

ffmpeg -i input.mp3 -af "volume='min(t/5, 1)':eval=frame" output.mp3

• t/5 increases the volume from 0 at \(t=0\) to 1 at \(t=5\).
• min(..., 1) ensures the volume caps at 100% (1.0) and does not continue to increase after 5 seconds.

3. Delayed Fade-Out

To keep the volume at 100% for the first 10 seconds, and then linearly fade out to silent over the next 5 seconds (ending at 15 seconds):

ffmpeg -i input.mp3 -af "volume='clip(1 - (t-10)/5, 0, 1)':eval=frame" output.mp3

• t-10 shifts the start of the fade to the 10-second mark.
• clip(..., 0, 1) restricts the output volume strictly between 0 (complete silence) and 1 (original volume), preventing negative volume values as time continues past 15 seconds.