Apple Music Radio Normalizes Music Volume with Sound Check

A friend at a classical music label sent me an email this morning, saying that he was listening to an Apple Music Radio station based on a classical artist, and that he was hearing some distortion in the music. When he listened to the same track on Apple Music, there was no distortion.

“Aha,” I said, “Have you not read this article on my website?”

It’s worth mentioning again: iTunes Radio (now Apple Music Radio) Normalizes Playback Volume with Sound Check. It uses Apple’s Sound Check feature, which is designed to make sure there’s not too much difference between the volumes of tracks on radio stations. This prevents jarring changes in volume, but it ignores the often large dynamic range (the difference between the softest and loudest parts) of classical music.

Here’s one example, from the older article, of Miley Cyrus’ Wrecking Ball. The first part of the waveform below is from the iTunes Store version of the track, and the second from the iTunes Radio version:



The quiet part of the song is much louder to match the overall loudness of the track, but also to be in line with the desired volume of the Apple Music Radio stations.

So be aware that when listening to Apple Music Radio, you’re not hearing the music unaltered; it can be louder or softer, and, for quiet music, there’s a good chance you’ll hear distortion.

2 thoughts on “Apple Music Radio Normalizes Music Volume with Sound Check

  1. Most *popular* radio stations use compression. Generally, classical music stations do not, and their overall sound levels are much closer to what’s on the original recording.

    On to the original post: If the quiet portions of the music that are distorted, it means that Apple’s Sound Check feature isn’t using normalization; it’s using dynamic range compression.

    Compression is a process by which the dynamic range of a recording is manipulated, as the track plays, to reduce the loudest sounds to a desired level, while also amplifying the quietest sounds to bring them up to that level or as close as possible. When performed on recordings with large dynamic range, the ‘pumping’ or ‘breathing’ of background noise in quiet passages becomes quite audible.

    Proper normalization requires scanning the entire track for the loudest sound level encountered. If that’s below some desired level (often 0 dB), the amount of amplification required to bring that sound to 0 dB is applied to the whole track. So the entire track becomes equally louder, and the overall dynamic range (the difference between the loudest and softest passages) remains intact.

Leave a Comment