Hi All,
For research purposes I am trying to assess and analyze audio from 4K Bluray sources.
Once I have the M2TS file successfully saved, when I import the file to Audacity and play it back (audio only) the peak loudness is way off, peaking over 11+dB above full scale, introducing significant clipping in those parts.
Having listened to the audio from the source, and from the M2TS via VLC this clipping doesn’t occur, so it is definitely from the way Audacity is handling this file.
Is there a setting I am missing, or something I need to do differently to have this work properly?
Measuring average and peak LUFS is part of my remit, so I want it to be as accurate and consistently repeatable as possible, thanks
It’s probably not really clipped. Some of these formats can go over 0dB without clipping (as can Audacity), but Audacity will “show red” for potential clipping. And if you play it at “full digital volume” you WILL clip your DAC. You can run Amplify at something like -15dB, then zoom-in to look at the top of the waveform to see if it’s clipped. (Then you can optionally run Amplify again to bring the peaks back-up to 0dB).
If you are downmixing from surround to stereo, mixing is done by summation so that will boost the levels, but again it shouldn’t actually be clipped (yet).
Dolby has a metadata field called Dialnorm and if FFmpeg is ignoring that, it will be louder. (I don’t know if the other Blu-Ray formats have something similar.)
Thanks for the suggestions. The Dolby Dialnorm for this track is -31 which is the standard, but I’m not sure what FFmpeg does with this. I know that dialnorms of more than -31, i.e. -23, are supposed to signal the processor to output a higher volume.
Audacity isnt identifying clipping in the individual channel waveforms, but when you play the audio in Audacity and all 8 channels play together, thats where the combined channels are clipping which is reflected in the main output scale and the LUFS readings (and to the ears).
I’ve also just tried with another file, this time a DTS HD 5.1 track and again - it’s peaking at +11.6dB true-peak-max and -4.7 Integrated LUFS.
Audacity or FFmpeg are importing the project at too high a volume, so the next question is “by how much” so I can apply a consistent reduction across all source files.
So I tried your trick of changing the amplify to -15 and that has improved the issue - but it’s too low now; we’re getting unclipped peaks of -5.3dB truemax, which is too low now - and when I want this to be accurate, I’m very uncomfortable with guesswork.
OK, so I tried a couple of other things.
I have extracted the TrueHD stream from the M2TS and tried that and same problem.
Converted the THD to 24 bit WAV and still the same issue.
I guess back to square 1 again.
OK so I tried the M2TS in Audition and it was pulling the descriptive audio and I wasnt sure how to get rid of that, but when I use the same WAV file that was going 10+dB over on Audacity and its not going above 0 on audition, so it must be an Audacity thing then?
