[FFmpeg-user] EBU R128 True Peak?
Boštjan Strojan
bostjan.strojan at gmail.com
Thu Apr 25 11:13:03 CEST 2013
On Wed, Apr 24, 2013 at 7:02 PM, Manu N <audionuma at gmail.com> wrote:
>> You have to keep in mind that the -1 dBTP limit is for PCM codecs. If your
> stream is to go thru lossy codecs, you should keep within a tighter limit,
> like -3 dBTP.
> So if there's a sort of warning, it should be a user setting more than a
> hardcoded value.
> Besides, depending on the sample format in a PCM stream (fixed vs float),
> raising the file1 to the target level might induce irreversible artifact on
> peaks (let's say you have a stream whose integrated loudness is -27 LUFS
> and max true-peak -3 dBTP, the +4 dB gain will create peakings if you are
> in 16 bits PCM).
Thats interesting, but do you actually have such stream in existance
or are you theoretically speaking? (Just did a quick scan on bunch of
"close to -23 LUFS" files and "sample peak" is max -6 dB, so basically
"true peak" would have to be 6 dB higher?
example sample:
documentary_trailer_short.wav
pcm_s16le ([1][0][0][0] / 0x0001), 48000 Hz, stereo, s16, 1536 kb/s - 25s 0ms
-9.0 dB max volume
-20.5 LUFS Integrated loudness (I)
2.7 LU Loudness Range (LRA)
-2.5 dB correction
-16 ++ + + + + ++
+ + + * + + +
-18 ++ * * * * * ++
* ** * * * ** * * * ** ** * *
-20 ++ ** ** ** * ***** *** * * * * * * ** ** ** ** ++
* * ** * ** ** * * * * * * * * * **
-22 ++ * ** * ** ** * ***** ** * ++
* * ** * ** * * ** *
-24 ++ * * * *** ++
* ** ***
* * ***
-26 ++* * *++
* * *
-28 ++* * *+
* * *
-30 ++* *+
* *
-32 ++* *+
* *
-34 ++* + + + + *+
+ * + + + + *+
0 5 10 15 20 25
Y Momentary Loudness (min=-35, max=auto dB) / X Time in seconds
B.
More information about the ffmpeg-user
mailing list