Reduce 24-bit or float sound to 16 bits with ffmpeg on Windows: plain conversion floors 24-bit…

Guide ID: art_01M7BWMH9XRRHCMXQYREEFMNY0

Editorial publication checks are separate from community and evidence review. The evidence labels retain this revision’s reported inspection, execution and review status.

Guide title

Reduce 24-bit or float sound to 16 bits with ffmpeg on Windows: plain conversion floors 24-bit and rounds float, what the dither methods change on a very quiet tone, and a script that measures the noise it adds

Coverage

For anyone who must deliver or import 16-bit sound from a 24-bit or float recording. A PowerShell script converts a WAV or FLAC with a dither you choose from ffmpeg's methods and reads the result back for length, level, sample peak, DC offset and how much noise was added in three frequency bands. It ran on generated tones, a fade and pink noise, with the plain commands, a scale setting and the refusal cases.

Example

A plain ffmpeg conversion of a 24-bit WAV to 16 bits moved every sample down by about half a step, while a float WAV of the same tone became silence. With triangular dither the tone kept its level and no offset appeared.

Useful starting points

• Name the dither yourself, in the same command that sets the 16-bit format; a plain conversion adds none.
• Read the added noise by frequency band, since shaped methods move it up and raise the total.
• Check the fade-out and room-tone tails by ear; the numbers show levels and shapes, not what you hear.