In my various sessions with Shotcut, I have noticed the reading on the Integrated LUFS metre is higher than what it is on other software plug ins I use. As I use audio a lot, this is beginning to bug me a little. So much so, I now have 2 metre readings when mastering the audio before going into Shotcut. The 2 individual meter readings are identical yet when in Shotcut, the LUFS meter reads the audio at a higher level than what is set in the mastering process.
Basically, is the Integrated LUFS metre in Shotcut up to date? Or can the 2 external individual and identical metre readings be wrong?
I think there are 3 timebases for LUFS (instantaneous, short window and entire programme) and a couple of weighting functions. Are you sure the meters are set to the measure the same parameters?
The Shotcut loudness scope runs in real time while you preview your project in Shotcut. How are you comparing that integrated tool with external tools? Can you give us more detail on the comparison that you are doing so that we can make the comparison ourselves to see the difference?
Basically, I use Audacity to master the audio tracks before going into Shotcut. Audacity is an open source sound editor wherein I use various plugins in the mastering process. Some of these plugins have integrated LUFS metres to monitor the output. During the mastering process, I am aiming for and achieving -14 LUFS integrated as this is a requirement for streaming platforms. This file is then saved and imported into Shotcut as an audio track but, the LUFS reading is different in Shotcut. A higher reading than what I get in Audacity with the plugins. Hence the question.
To calculate the integrated LUFS in Shotcut, the playhead needs to be rewound to the beginning of the video, then the Reset button on the LUFS meter clicked, then the video played from start to finish with no skipping forwards or backwards. For a 5-minute video, this means waiting 5 minutes to get an integrated reading (“integrated” by definition meaning the entire program). Is this the measuring methodology you are using? If not, that probably accounts for the difference. Many people either forget to click the Reset button which conflates readings from all clips played since the last reset, or people don’t run the LUFS meter over the entire program when comparing integrated readings.
I did two quick tests using Shotcut 23.06.14. I did Open Other > Audio Tone and put a 1 kHz tone at -20 dB in the source viewer, then reset the LUFS meter and watched the scopes. They sat perfectly on -20 LUFS. Next, I got a -20 dBFS RMS uncorrelated pink noise sample from Bob Katz’s Digital Domain site (requires a login), reset the LUFS meters, and watched the scopes. The values were around -19.3 LUFS, which is about as good as expected on pink noise (since integrated loudness on pink noise always depends on the length of the clip being metered, “pink” being a random distribution and all).
At quick glance, the LUFS meter looks good. If you find otherwise, I’m happy to analyze an audio clip if you upload it here.
You did not say what tool in Audacity you use to measure loudness. So as a comparison, I downloaded the free tool from this site: METER — Orban
I opened a 20 second video file and analyzed with the orban meter. The result is -26.5
Next, I followed @Austin’s instructions to use Shotcut: start a the beginning of the project, reset the meter, and play the entire project from beginning to end. After the preview player got to the end and stopped, the integrated loudness (I) bar showed -26.5.
So at least Shotcut agrees with that one free tool I found. Maybe you can follow the same steps with your tool - or give us more information about your tools so we can try it ourselves.
Yep, I reset the loudness metre every time I rewind the video or the audio. It is a routine I go through each time as it is the correct method.
The 2 plug ins I use are plugins that are paid for hence why I didn’t mention them. I didn’t think you would buy them to try them. (Waves loudness metre Loudness Meter Plugin – WLM Plus | Waves and TDR Limiter 6 GE | Tokyo Dawn Records are the plug ins I used.
I did some testing on this today using the official loudness test set files:
The expected loudness measurements are listed on page 10 of this document:
In my testing, the loudness meter is giving the expected results UNLESS the Shotcut project Audio Channels setting does not match the number of channels in the source file.
Can you check your project settings and verify that your project channels setting matches the source file channels?
Also, can you do some testing with the test files I linked above?
Side Note: The loudness measurement is not expected to match the reference level if the channels are downmixed before being re-measured. As a simple example, consider a stereo source file normalized to -23LUFS. If you convert that source file to mono by adding the samples and dividing by two, and then re-measure it, the measurement will come out to -26LUFS. The reason is that the measurement algorithm calculates that two speakers are twice as loud as one speaker.