The GDELT Project

Using FFMPEG To Measure Audio Levels On Television News

What audio levels do television news channels use? There are long-established standards for television production, but what levels are observed by a typical home consumer? Using FFMPEG's "volumedetect" feature we can get some estimates using:

ffmpeg -i ./VIDEO.mp4 -filter:a volumedetect -f null /dev/null

Here are the results for a randomly selected CNN broadcast from Dec. 1, 2020 ("volumedetect" only displays the first few entries of the histogram):

n_samples: 351303680
mean_volume: -32.3 dB
max_volume: -4.7 dB
histogram_4db: 13
histogram_5db: 16
histogram_6db: 27
histogram_7db: 73
histogram_8db: 272
histogram_9db: 668
histogram_10db: 1558
histogram_11db: 3899
histogram_12db: 7676
histogram_13db: 16404
histogram_14db: 35266
histogram_15db: 69792
histogram_16db: 130424
histogram_17db: 232799

The results for a BBC News broadcast the same day:

n_samples: 178550784
mean_volume: -26.4 dB
max_volume: -5.5 dB
histogram_5db: 6
histogram_6db: 62
histogram_7db: 799
histogram_8db: 5850
histogram_9db: 29804
histogram_10db: 83199
histogram_11db: 177939

The results for a Fox News broadcast that day:

n_samples: 351348736
mean_volume: -31.9 dB
max_volume: -7.3 dB
histogram_7db: 2
histogram_8db: 62
histogram_9db: 233
histogram_10db: 710
histogram_11db: 1474
histogram_12db: 3155
histogram_13db: 8520
histogram_14db: 19879
histogram_15db: 42304
histogram_16db: 86969
histogram_17db: 167447
histogram_18db: 309361

The results for an MSNBC broadcast that day:

n_samples: 351356928
mean_volume: -32.8 dB
max_volume: -8.3 dB
histogram_8db: 25
histogram_9db: 35
histogram_10db: 68
histogram_11db: 243
histogram_12db: 662
histogram_13db: 2318
histogram_14db: 5942
histogram_15db: 14843
histogram_16db: 33365
histogram_17db: 66772
histogram_18db: 135605
histogram_19db: 297055

All three US channels are fairly in sync with one another, while BBC News is slightly louder, though it isn't clear whether that is just the one randomly selected broadcast or a more consistent pattern.