OPUS must be some kind of miracle codec if 128kbps is appropriate

thanks I've been waiting for this "Gi(a)rlish Number" when ?

@motbob it is. Here's a graphic from the [hydrogenaudio 96 kbit/s listening test](http://listening-test.coresv.net/results.htm) 

The most interesting thing about that test to me is the tendency of testers to rate 128kbps MP3 as "not annoying." But I find it really annoying. Am I... an audio snob? Oh god, I need to reevaluate everything.

Really depends on the content. For voice, anything better than 64k HE-AAC sounds decent to my ears, but that's not the case for music. This should be just fine.

@aryma_saga I just needed a timer and finally got one. He said he probably needs a few weeks though. It's definitly not dropped!

@motbob And yes, it is. You don't need that much bitrate with proper AAC as well (see below). For anime 128kbps-160kbps are fine. The quality is almost the same as OPUS (OPUS is just slightly better and smaller). The problem I have with AAC are the multiple encoders. Nero and the native FFmpeg AAC are bad. qAAC and FDK AAC good. Most encoders don't tell which one they use. For OPUS only one encoder exists. So you can always be safe here.

what kind of madman doesn't use qaac or fdkaac

Probably Doki for example. At least their AAC audio sounds pretty terrible, even with ~130kbps.

I definitely agree with the AAC encoder thing, I try to remember to put QAAC in my torrent descriptions because sometimes I download AAC with ffmpeg encoding and it makes me really mad and wary of AAC (COUGH SOME-STUFFS COUGH)

>doki

haha more like dookie

DTS HDMA / FLAC is the best route for me :^)

Same here, aryma_sama. But I guess some people have limited storage requirements.

well DTS core is enough for a movie / series . but i'll choose dts-ma / LPCM for concert video or some anime like *cough*lovelive&im@s*cough*

>using uncompressed audio

You sure OPUS isn’t gonna just turn into another Vorbis? Also qAAC should be the only choice when choosing this file type no exceptions. @motbob Nah if you own a decent ($110+) headphone like Audio‑Technica, Sennheiser and so on maybe with an amp too.

OPUS is still smaller and better than qAAC. The compatibility nowadays is pretty good. It's used with VP9 and soon with AV1.Opus is the audio of the future, so I prefer that.

eXmendiC, i like that side of you, prefering new codecs over "safe" ones, but i'm curious why you're not that way when it comes to x265 ? is it still unmature in your opinion ? even when a lot os encoders trust it these days

From: https://trac.ffmpeg.org/wiki/Encode/HighQualityAudio ``` libopus > libvorbis >= libfdk_aac > aac > libmp3lame >= eac3/ac3 > libtwolame > vorbis > mp2 > wmav2/wmav1 For AAC-LC: libfdk_aac > aac NOTE: as of 2017 this is no longer necessarily the case. The built in aac encoder is quite good. The >= sign means greater or the same quality. ``` The native ffmpeg aac audio encoder, historically, had really bad handling of VBR audio, but CBR was okay. Now it's native CBR is roughly the same as fdk_aac. If you really care about audio quality, you should insist on FLAC/lossless or Opus at very high bitrates (192k+), not worrying about which exactly which encoder was used for a given inferior codec (AAC).

>why you’re not that way when it comes to x265 x265 is much slower and still worse. It's not usable for normal encodes at the moment. Psy is pretty fucked up and creates you banding while encoding. I wouldn't trust any of these encoders tbh, but I haven't seen a decent group using x265 (except as a meme).

@YukinoAi 192kbps OPUS is overkill for anime. 160kbps is the max you should use. 128kbps the best match between quality and file size.

Also another problem with x265 is that you have to learn much new stuff. It’s not as easy as x264 (imo at least). You’ll rarely see any good h265 encodes (because of some bugs and that most encoders have no idea how to properly use x265).

ok thank you for replying, i'm using avisynth with x264, but i don't why no matter what i do ican't get the dither options to work for me, always giving me error messages, loaded different ver of dither and other dlls that it needs along with different avisynth vers, i used your guide first before doing more tries, but still no luck

The audio bitrates just do not matter much to the final file-size so there is not any point in being frugal with the audio (128kbps vs 192k) bitrates. If you are wondering what the best acodec is between 100k-140k, then it does not make sense to decline to do the same with the video because the video represents such a large % of the final file-size.

At ~800MB/epi, that is ~<30MB for the audio or ~3-4% of the overall size. The video is the other __96%__. The audio size is literally masked within the variance for the video stream sizes. If you don't care about file-sizes and a lot about quality, just use FLAC and be done with it. If you do care about file-sizes, x265 is the way to go. That can take the video/audio % to about ~70% for multi-audio files. That said, you are right about x265. It is very prone to artifacting compared to most AVC encoders (esp x264). There is also no actual benefit to it besides lower file-sizes which are only possible at comparable quality by letting x265 take as long as it needs, which can be a very long time. Medium/fast preset h265 vids, or worse (gpu encoders) are just bad. It results can be [quite nice](//diff.pics/FlgrP5kfjiXt/1) however.

@Chilled No idea, I don't use Avisynth. I use Vapoursynth. @YukinoAi I don't hear any differences between 160 and 192. Between 128 and 160 barely any. However, for new shows I use 160. I care about quality and file size, so x264 with Opus is the way to go for now (lossless audio is mainly bloat). I am doing the same with the audio as to the video (except filtering) - Good compression with low file sizes.

I really don't get it. Why would you cheap out on the audio? Your encodes are not exactly small and Opus@128 saves you like 20mb compared to qaac -V 127, which is objectively better.

Because I started that project like half a year ago and don't change the 128 to 160 in the middle of the show? I wouldn't even use qaac -V 127 when -V 91 literally sounds the same with less file size and is equal to opus with 160kbps (which means I still would use opus).

Maybe 160kbps opus is actually better than qaac -v 91, I haven’t made any comparisons. However, it’s not “cheaping on the audio”. I just see no benefit for higher file sizes without any noticable difference. It’s like using crf=12 instead of crf=14, because you see difference only in still frames and not while watching. There is a limit for everyone and mine (with any new projects) is 160kbps with OPUS and crf=14 (720p) / crf=15 (1080p).

People can always mux flac/pcm/aac or whatever into my encodes. The good thing is that you just have to trim 24 frames for BDs (on some BDs you don't even have to trim anything). That makes it pretty easy muxing another audio into it. However, chances are pretty low you'll see aac/flac for my projects (except I'm encoding it for another fansubgroup).

i'm with you, i can't hear the difference between 128 and 160, but other people, the reason is simple no audio setup is equal to another...

@Chilled http://avisynth.nl/index.php/Dither_tools That will tell you what you need. Use avspmod (if you don’t already got it) and load in the scripts Ex: LoadPlugin(" path \dither.dll") See http://avisynth.nl/index.php/Dither_tools#DitherPost of what dither to use under mode.