wat. please tell me you're losslessly cutting the english audio to make it fit.

>please tell me you’re losslessly cutting the english audio to make it fit. lol

oh no

When a group trims HDTV audio to fit to a web video source, is that lossless cutting if the original AAC audio track is lossy? No, but they do it anyway. We used that same method, so if you guys have a problem with that, then a lot of trusted groups have been screwing up for a while now.

By "trimming," you mean something using ffmpeg -c copy or something similar, yes? I just want to make sure we're on the same page.

Yes, precisely. We upsampled the audio for episode 02 which honestly doesn't really change anything (trust me, I've tested it; no audio degradation happens in this case... the English audio is already pretty shit as it is unfortunately). But someone requested it in a comment, so why not? Hopefully US BDs come out soon so that we can get better audio overall for the dubs.

>When a group trims HDTV audio to fit to a web video source, is that lossless cutting if the original AAC audio track is lossy? why would it not be lossless if the audio is exactly the same afterwards? also no part of the process involves the sample rate so i'm really curious how that is involved.

It's possible to losslessly trim audio. It's impossible to losslessly trim audio if you're changing the sample rate--you have to reencode to do that. I promise you that "a lot of trusted groups" are not reencoding their audio.

...but re-encoding web videos is okay? I really don't understand this community. Its almost like you guys pick and choose who to ridicule for things. If it does nothing to the audio's quality like I said before, why is it such a big deal to you guys? Someone asked for it, so we got it done because we want to help keep the people who download our stuff satisfied.

we re-encode the video to improve the quality. you can't do that with lossy audio. in fact, re-encoding it tends to lower the quality. besides, you're missing the point. you don't need to re-encode the audio for any reason.

>When a group trims HDTV audio to fit to a web video source, is that lossless cutting if the original AAC audio track is lossy? No, but they do it anyway. We used that same method, so if you guys have a problem with that, then a lot of trusted groups have been screwing up for a while now. Why would it not be losslessly cutting? It does no re-encoding. What exactly did you use for this? People who trim tv audio and sync to web sync the two video sources by either using multiple trims or a single trim then syncing the two video sources up. vfr.py is used for cutting/trimming which just uses mkvtoolnix. That literally can't re-encode. It doesn't have a built-in encoder.

Sample rate conversion does not lower the quality of an audio track. It is a common technique used to convert compact discs to digital audio tapes, for example. https://en.wikipedia.org/wiki/Sample-rate_conversion

Well, I'm not trying to ridicule you. But reencoding audio is something that will definitely lead to a worse user experience for some people. You might choose a too-low bitrate or use a bad encoder. Even if you can't notice the difference, that doesn't end the issue because different people have different sensitivities to bad audio. For example, SmugCat went half a season of Goblin Slayer without noticing that their audio was totally broken. EDIT: It looks like you're trying to defend what you're doing as correct. You should stop and just find someone you trust who knows a lot about audio if you don't trust us.

>Sample rate conversion does not lower the quality of an audio track. it does when the source is lossy since it necessitates re-encoding. besides, you still haven't explained why it was necessary.

None of you have said any reason as to why sample rate conversion is bad except "we don't do this because its bad." I use vfr.py just like everyone else to trim it, except I converted the sample rate first. If it was used just as that Wikipedia article says to convert 44.1 kHz CDs to 48 kHz digital tapes, then I'm going to come to the conclusion that there is no problem with converting sample rates.

Sample rate conversion is not a lossless process, and in fact quality varies from program to program. You can compare the results from several software packages and the artifacts they introduce here: http://src.infinitewave.ca/

>Sample rate conversion does not lower the quality of an audio track This is true, sometimes. BUT SRC is a lossy process. And you're reencoding, which is ALSO lossy. Going 44>48k wont yield many benefits. JUst because you can't hear or see the difference doesn't mean there is none. Just use mkvnix if you can (vfr.py is a frontend for that). ED: wiki has 0 context on anything. The CD > DAT example is VERY OUTDATED and noone uses DAT for audio. Studios nowdays record at a SR, and kepp it that way for the entire chain if feasible.

even if it's not bad to do, there isn't any reason to. it won't magically make the audio sound any better.

We did it 100% for consistency with the Japanese audio's sample rate, but since what seems to be a majority of people seem to not like the idea so much, we are open to reverting it and keeping the English audio at 44.1 kHz.

> english audio

44/48k consistency you won't notice unless you have a very broken playback system.

@nks: Trust me bro, I don't watch dubs either. I'm just a nice person and want everyone to experience the Mongolian 2D puppetry in the best quality they can. xD

what i would do in this case is take the lossless japanese audio from the japanese BDs, convert it to 44.1khz, encode it at whatever bitrate AAC like the adult swim audio, and then splice it in for the OP and ED. i think that would generally produce the best result that doesn't lower the quality any more than necessary if you really don't want to wait for the english BD release for the dub.

Thanks for the advice, I'll make the correct changes based on everyone's feedback in the comments.

@MTBB > Even if you can’t notice the difference, that doesn’t end the issue because different people have different sensitivities to bad audio. For example, SmugCat went half a season of Goblin Slayer without noticing that their audio was totally broken. If by “totally broken” you mean lower quality, sure. Keep in mind that despite hundreds of downloads every week, you were the first and only person to report a problem with it. I did not ever anticipate the audio to be of such low quality as the source was a WEB-DL, which I figured would have untouched audio, the same way the video was untouched. Additionally, the size and bitrate of the audio turned out to be virtually the same as the proper untouched Wakanim tracks. I had no reason to look at spectogram beforehand and never ended up watching the show, which is why I didn’t notice. Kuromii QC’d everything but she’s not very tech-savvy and she mainly focused on the subtitles, so I can’t fault her for not noticing either.

IANAE but being "lossless" means that no data is virtually lost, and the proof for it is its ability to reproduce the original source data bit by bit (i.e. reversible). So by that, any re-encoding/resampling that isn't reversible is a lossy conversion, for the original data is lost and can't be reversely retrieved using the output anymore. I'm not against any re-encoding, but I guess we should be stricter on defining which ones are lossless and which ones are re-encodes. We can call a converted output as something like an improved version, and it may really is one. But that should still be defined as "lossy" regardless if it's better or worse than its input source, since it already "lost" the data in its original form that it was published.

@LastReaction was this discussion because I requested that ep.2 change? I'M SO SORRY! It was just my OCD acting up, I wasn't complaining! I love and respect you guys for doing this kind of work, even tho nobody asks you. You're above KiritoJesus in my praise list

>words Scyrous, you spent a lot of time defending yourself without realizing that I already did the defending for you when I said "different people have different sensitivities to bad audio."

@MTBB - I think you misunderstand. I totally agree with you that different people have different sensitivies - no doubt about that. I'm just saying your comment doesn't apply in my situation because I never listened to the audio in the first place.