Hello,
I am musician and making some videos with my music. Problem is that audio which I am importing in kdenlive (newest version) is always +2dBs louder then it is in Ableton DAW, for example. I checked effects, project settings and there is nothing on like normalisation of audio. I know that in kdlive’s mixer can go -2dBs with it but why is that happening? Need help here?
Btw.software is amazing!
Hi, and welcome to the forum and community.
Excuse my stupid question, but how do you know that the volume is up by 2dB?
Silly question #2 (: Are you sure it’s ‘kdenlive’ boosting it and not something in the Ableton pipeline attenuating the volume?
I’m assuming you mean dBfs here? Can you run a test against a reference signal to see if either or both change the level relative to it. And what is your source material? It’s not impossible this is happening at the decoder level, or possibly in conversion from float to int samples (though 2dB seems a bit much for that), or decoding multichannel audio to stereo, or similar.
kdenlive itself doesn’t really do any audio processing, all of that is handled by external libs, so the game is going to be pinning down where in the pipeline something different from Ableton happens, but +2dBfs sounds suspiciously like naively mixing in extra channels, and kdenlive support for multichannel audio (beyond stereo) is still work in progress.
Hello, thanks. I have measurement tools for that as I am mix engineer too 
Hello, I mean yes dB full scale. Yes I made track in ableton and have it on my spectogram on -0.5 dBfs, when I import it in audio channel on kdenlive it is going to red. Then when I make it quiter on mixer in kdenlive for 2dBs I got same loudness on my spectogram for it. It is same for example on mediaplayer on windows vs kdenlive, like ableton vs kdenlive.
I found different types of using my interface audio drivers in kdenlive but its not helping at all, same thing.
Yeah, there’s lots of different places this could be getting messed up - aside from the different driver interfaces, most DE volume controls have have ‘per-application’ volume setttings these days too, so you need some independent way of verifying the base levels.
What’s your spectrogram tapped into? Is it analysing a generated file directly, or the output from your audio hardware? And are you only seeing this when using kdenlive for playback, or in the rendered output file too?
First thing I’d check is to generate a file with Ableton, import it to kdenlive and render it, then compare those two files (analysing the raw data, not the playback volume produced by actually playing through the hardware) and look for any discrepancy. If you still see the ~2dB difference, then I would feed the file kdenlive rendered back into kdenlive again (import it into a new project) and render it again, to see if the result is ~4dB different from your original …
That should put some rings around what we can rule out and where we should look next.
What version of kdenlive are you using and on what platform? Is this all on Windows? And if so, are you able to run some tests on Linux too? (if Kdenlive really is boosting the sound level every time you roundtrip a decode/encode, it would be good to know if this is Windows-specific or happening on Linux too)
Also what codec are you using in what container? Is this raw PCM, or some compressed format?
Ok, went more deeper into this problem and it was Kdenlive vs Sonarworks app. So I need to go down when listening in sonarworks then to 0 dBfs when rendering. I matched audios from rendered kdnelive with audios in Ableton in LUFS. Difference is -0.1 LUFS which is added to kdenlive and its converters to mp3.
Another question, is there option to convert audio in .wav file with video when rendering? I am importing .wav file from Ableton to Kdenlive and dont wanna loose sound quality.
Thanks and sorry for my brief overview for first question.
Cool, that’s going to be ‘in the noise’, either quantisation if you really used mp3, or even slight phase differences in the transcoded signals. Perceptually they’d be ‘identical’.
Another question, is there option to convert audio in .wav file with video when rendering
From the render dialog you can create your own custom profiles for codec/container options if none of the preset ones fit what you need - but for a ‘final’ render (that is intended for viewing, not further editing), a good audio codec like Opus will be transparent (perceptually lossless) even at relatively low bitrates, so I’d still use that even if you want the highest quality audio.
Thanks and sorry for my brief overview for first question.
You’re welcome and no worries! You never quite know who is at the other end and what their experience might be until you start a conversation - but if you do see anything else like this don’t hesitate to bring it up for investigation, some audio bugs can be subtle and hard to spot, so when someone who knows what they are looking at sees something weird it’s always worth trying to get to the bottom of it.