Tweet
Hi, first I want to say - great product!
I've read a few forum posts asking similar questions about stereo to mono conversion, and the advice is always to convert to 32 bit floating, use channel mapper to convert the 2 channels to mono, volume normalize (peak to peak) and then truncate (with or without dither) to 16 bit. But I want to skip the normalize part, and have the peak level the same as the original stereo file. While I think I understand this a little, please correct me if I've got it wrong...
The loudest possible part of the new mono file will be where the two channels of the stereo file have their waves both at 0dBFS, at the same time, any other points within the file where the waves are summed will be lower than that, that is the highest it could possibly be. Furthermore, the resultant mono version then will have it's highest theoretical peak at +6dBFS (floating point). Now, I say this because I think I understand that amplitude doubles with every 6dBs, is that right? That means the high peaks have combined, doubled, therefore the new peak is +6dBFS. Also, from what I understand a 32 bit floating file can accommodate that kind of digital "over", is that also correct?
So, to get the result I seek, I need to convert to 32 bit, sum to mono, reduce by 6dBs, then truncate at 16 bit (with or without dither). Would that get me a file with the same peak level as the original, but now mono? If that is correct, how do I reduce the volume by 6dBs? Fixed amplification won't do it, because setting that at -6dBs will make all files peak at -6dBs, and that's not what I want either. I want each file to peak the same as its original mono file.
To put this in context, the actual task I am attempting is to batch convert 10,000 stereo drum hit samples (kicks, snares etc) that I need in mono, but their current peak volumes (relative to each other) needs to be maintained, I don't want them all normalized. Some are already peaking at 0dBFS, but some are no where near that, and that's fine... what's important is that I maintain that relativity.
Thanks for your help!
ps. I also considered summing each channel at 50% of it's original, which means again that the highest peaks, when combined, will be brought back up to 0dBFS. Is this a more elegant solution? Which one will get the best result?
I've read a few forum posts asking similar questions about stereo to mono conversion, and the advice is always to convert to 32 bit floating, use channel mapper to convert the 2 channels to mono, volume normalize (peak to peak) and then truncate (with or without dither) to 16 bit. But I want to skip the normalize part, and have the peak level the same as the original stereo file. While I think I understand this a little, please correct me if I've got it wrong...
The loudest possible part of the new mono file will be where the two channels of the stereo file have their waves both at 0dBFS, at the same time, any other points within the file where the waves are summed will be lower than that, that is the highest it could possibly be. Furthermore, the resultant mono version then will have it's highest theoretical peak at +6dBFS (floating point). Now, I say this because I think I understand that amplitude doubles with every 6dBs, is that right? That means the high peaks have combined, doubled, therefore the new peak is +6dBFS. Also, from what I understand a 32 bit floating file can accommodate that kind of digital "over", is that also correct?
So, to get the result I seek, I need to convert to 32 bit, sum to mono, reduce by 6dBs, then truncate at 16 bit (with or without dither). Would that get me a file with the same peak level as the original, but now mono? If that is correct, how do I reduce the volume by 6dBs? Fixed amplification won't do it, because setting that at -6dBs will make all files peak at -6dBs, and that's not what I want either. I want each file to peak the same as its original mono file.
To put this in context, the actual task I am attempting is to batch convert 10,000 stereo drum hit samples (kicks, snares etc) that I need in mono, but their current peak volumes (relative to each other) needs to be maintained, I don't want them all normalized. Some are already peaking at 0dBFS, but some are no where near that, and that's fine... what's important is that I maintain that relativity.
Thanks for your help!
ps. I also considered summing each channel at 50% of it's original, which means again that the highest peaks, when combined, will be brought back up to 0dBFS. Is this a more elegant solution? Which one will get the best result?
Comment