I'm starting to really like Handbrake as a tool for conversion, as well as for ripping of both DVD and Blu-Ray. While it doesn't have support for all of the output formats I would like (would like to still be able to use AVI H.263 XviD output if I could), I find I'm liking the H.264 MP4 formats better and better, even though my XBox 360 won't play them with 5.1 audio (stupid Microsoft). However, I'm trying to learn my way around some of the more advanced settings options, and I have a quick question... What are good settings for the DRC option to cause strong compression, while still maintaining enough DR for impactful nuances in the audio? Also, a probably not so quick question... how does Handbrake implement DRC, and how does the parameter provided impact that implementation? Following are more details.
Ordinarily, I find that most movies these days have really loud sound effects (and often really loud scores), while having somewhat quiet or really quiet dialog. I understand the reasoning for this, and in a home theatre environment when you don't have to worry about disturbing the neighbors or other people living in the home, it can be a wonderful audio experience for the most part. But I do have neighbors to worry about, and other people in the home who have very different work and sleep schedules. And in addition to that, there are some movies that just take the audio dynamic range way too far, so that when you have the volume loud enough to actually hear the quieter dialog at all, the louder stuff, like explosions and gunfire, and music, is often painfully loud (literally). Because of all of this, I have made it a habit to use DRC on pretty much all of my rips and conversions.
Normally I would load either a video file or an extracted audio file into Audacity and use the Compressor filter to inflict DRC, then remux. But if I can use Handbrake to perform DRC on the fly while converting or ripping in the first place, then I'd rather do that. I understand the limitations of Handbrake DRC in that it will only work with AC3 audio streams, and only when performing audio conversion, not passthrough (that seemed obvious to me, but I guess they felt the need to make a point of it in the help file), but a large number of movies use AC3 audio these days anyway.
Because I'm used to using Audacity for DRC, I'm used to advanced full-featured configuration of the compressor filter. The settings I most frequently use result in some pretty significant compression of the DR, and are as follows...
- Threshold: -30.0 dB
- Noise Floor: -50.0 dB
- Ratio: 4:1
- Attack Time: 0.3 sec.
- Release Time: 3.0 sec.
- "Make-up gain for 0dB after compressing": enabled
For my purposes I find this to be nearly perfect in almost all cases, as it results in strong DRC while maintaining enough nuance for me to notice differences between, say, a whisper and normal dialog, or between soft music vs. the peak of a crescendo.
I'd like to be able to achieve similar results from Handbrake's integrated DRC, but with having only one parameter to set, I need to know more about how that parameter actually works and interacts with the DRC algorithm used by Handbrake. The help documentation states that the DRC will raise the audio of quieter signals to be louder and closer to the loudest peaks, without changing the peaks. Does this mean that they use some form of peak gain instead of full DRC? Is the parameter simply a setting of how many dB you want the quieter signal peaks to gain? Or do they use some sort of standard values for most settings, and the parameter is to specify a ratio, such that entering a parameter of 2.0 would result in a 2:1 ratio? (Though if that is the case, then why does it use a floating point)? Or is the system actually much more complex than that, and the documentation merely dumbed it down a lot for average users?
So I'd like some more information on how Handbrake actually implements DRC, and how the entered DRC parameter affects that implementation. But I'd also like some suggestions for some parameter settings that might be just what I'm looking for in terms of the strength of the compression.
You help is appreciated,
+ Reply to Thread
Results 1 to 6 of 6
Last edited by Drake1132; 20th Jun 2014 at 12:33.
OK, never mind... I had posted this simultaneously to the Handbrake forums, being unsure of where I would get the most or fastest help. I received an excellent reply over there almost right away. If anyone else has the same question and found their way here, please view that thread at the Handbrake forums. Hopefully others will find it useful
Edit: I posted this before seeing your second post, in case you're wondering why the conversation seems to be a little out of order..... but you might find some of it useful.
I'm not going to be much help answering your questions as I don't use Handbrake, but I do use the "turn up the quiet parts" method of volume levelling myself, and I think it tends to work better than compressing. I don't encode the audio that way as I use my PC for playback so I leave the audio as-is and "compress" it on playback. The DSP I use lets you set the maximum amplification, and the level up to which maximum amplification takes place. Above that, the amplification is "linearly interpolated up to the maximum level".
The upshot of it though, is the same applies as it does to compressing.... overdo it and you'll hear the "pumping" of background sounds at times etc, but because the DSP works by turning stuff up, there seems to be less need to adjust the threshold and compression level etc for each individual track.
Sorry I can't actually help with Handbrake, but until someone who can comes along, here's the thread where I explained my "compression" method if you're interested (posts 6 & 15). If you decide to give "turning up the quiet parts" a shot, and you happen to compare it to your usual method, I'd be interested to find out if you think there's much difference between the two. I can't say I've applied the usual type of compression to soundtrack audio much.... well possibly never.... so I've not compared them myself. And I've no idea how "clever" Handbrake's DRC implementation might be......
Last edited by hello_hello; 20th Jun 2014 at 12:52.
I don't even know how AC3 DRC works exactly, or how it'd compare to applying compression "manually" when encoding. Anyone know much about AC3's DRC?
I'm by no means an expert, but the way I understand it is that when audio is professionally encoded to AC3 format using the Sony authorized professional encoder(s), then the DVD/Blu-Ray authors can choose to enable DRC on the audio stream. What information I have indicates that doing this allows the authors to apply DRC of varying levels and settings to specific time sequences in the audio stream. So they could, for example, enable DRC for time 00:13:48:217 through 00:14:27:581 and then separately enable DRC with different settings to time sequence 01:03:51:003 through 01:22:14:530. Then, if that disc is then played in a device that is capable of reading and applying AC3 DRC levels, the volume for the output from the device would be adjusted on the fly based on the DRC flags present in the AC3 file. But if the disc is played in any other device, then the DRC settings are ignored and no adjustments are made.
To a certain extent this makes some sense. Doing it this way could allow a disc author to adjust audio from the original digital master that had audio set up with the intention of theater environments, and they could selectively tweak it to be more useful for a private, in-home viewing environment. But with it being so restricted in actual output usefulness, I don't really see the point in authors making that sort of effort.
As for the way Handbrake currently works, based on the info I got in the other forum, You can set DRC to "1" to enable baseline DRC based on the AC3 DRC flags, or you can make it more intense or less intense by using float values that are !=0 and !=1.
On a separate note though, I discovered today that using Handbrake to re-encode from MKV to MP4 and converting audio to AAC sometimes (possibly always) automatically remaps the audio channels. Not sure why it does that.
Over the past month I'm starting to like XMedia Recode more and more. It seems pretty stable for the uses I've put it to, and allows lots of advanced settings tweaking. It won't do anything for DRC, but it can be used for "normalization" using Replay Gain algorithms (which often naturally tends to adjust DRC levels a little in my experience).