Overmodulation when using ReplayGain
Reply #18 – 2012-10-20 00:26:17
Yes, thats why I ask to be sure if I've understood it right ;) @shakey_snake: Sorry for my complicated writing. I just wanted to know if the reason for reducing volume is to increase the dynamic. That is what I read in the link. The reason for the 89 dB SPL target that often reduces volume is to leave the dynamic range as it is, not to increase it or reduce it. The loud albums produced today are about 10 dB above the standard Replay Gain level and have a very consistent loudness which is about as loud as it is possible to make sound. They can't contain dramatic moments with increased loudness, because there's nowhere louder to go. You cannot guess which parts were supposed to be louder before this was mastered into the music, so there is no magic DSP to improve these loud and undynamic albums. If you were instead to increase albums from, say 1990 (or classical music) to the same loudness as today's loud* music, you'd have to clip it or limit it severely and stop the loud dramatic moments from being loud and dramatic in the process. You'd make them sound bad or at best bland. * The 'loud' music often sounds quiet or at least flat when played, because you own the volume control and it can contain no louder-than-normal moments that stand out. Instead, you should set a level more typical of 1990 CDs that allows the headroom for dramatic peaks, turn down the volume of today's music to match and then turn up the volume control on your amplifier to bring it back to your desired loudness. If you use Album Gain, the intentional differences between tracks in an album are preserved, but you can play an album from 1990 next to an album from 2012 and they will be at the same loudness (and you can shuffle tracks from many albums too). Consistent loudness should be the prime aim of using Replay Gain unless your audio hardware is volume-crippled and you're forced to compromise and work around that. For that reason I'd suggest using Album Gain and Apply Gain (without clipping prevention), and in your DSPs put the last DSP item as Advanced Limiter (which does nothing to your audio unless clipping would be about to occur, when it gently limits peaks rather than harshly clipping them). If you turn down fb2k's Volume control slider, Advanced Limiter is even less likely to do anything as you're providing even more peak headroom. This protects against nasty clipping distortion and protects against the incorrect volume that "Apply Gain with clipping prevention" mode would cause for the sake of avoiding a few milliseconds of rare clipping. If you set Pre Amp to 0 dB (89 dB SPL) for files with Replay Gain information and perhaps to -7 dB to -10 dB for those without, things should be fairly consistent. I usually use Windows Mixer (on Windows 7) to set the volume of system sounds to about 30% so they're reasonably consistent with ReplayGain-aware applications. You could also put other non-ReplayGain aware items (e.g. web browser, Flash Player, Skype etc) at similar volume in the Windows Mixer. The Windows master volume control then keeps the relative settings while varying the overall volume. If your sound system is good, you could go further into high quality headroom by setting pre amp to -6 dB (83 dB SPL target) and increasing amplifier volume accordingly. This makes it even less likely that the Advanced Limiter will need to step in and limit peaks that would clip by allowing even more headroom. Noise floor should not remotely become a problem even if you only have 16-bit audio and like to listen very loud unless your soundcard is very poor. If you have 24-bit output, every detail in your CDs, including the noise floor can reach the soundcard and amplifier. I actually have a different User Account named 'Music' on my Windows laptop where I adopt an 83 to 86 dB SPL target and usually mute my system sounds, rather than 89 dB SPL with system sounds on the account that I use every day. I use that account when plugged into the Bose Tonematch system used on-stage by the group I prepare backing tracks for. We can then edit with plenty of headroom for top quality and export 16-bit Apple Lossless files, fine-tuned for desired volume (including some deliberately louder than Track Gain might suggest), for use on the group's iPad which sits on a music stand. The iPad is persuaded to play one backing track at a time then stop by putting each track in a separate playlist, and interval music can go in another playlist that can be faded then stopped before starting the second half of the set. The analogue or digital noise floor of 16-bit playback via my onboard soundcard has never been a problem with these fb2k target levels, even with the Tonematch channel at high volume in the quiet of the practice room.