I understand your line of thinking that if you output a video, most people lay on the audio after. This is true, but you are missing a big point…in having the ability to place audio on an object or area you are creating a great binaural audio track that can’t be effectively duplicated in post. The fact that the audio is created so it can be heard realistically via the camera, the binaural audio track is something that could be used in wonderfully new ways.
Please adjust your thinking about this and add the ability to output a high quality binaural audio track in the sequencer!
If you have recorded binaural audio for your media, you can just play it as a non-spatialized stereo sound source, which will allow it to retain its binaural coherency. This should work whether in sequencer or not.
Understood. I do think you should add the ability to allow the sequencer to output audio however. Having to capture the audio separately, then synch afterwards in another program is silly. You already have the audio, it can’t be impossible to take that and apply some codec/encoder to be output along with the video. The sequencer is basically an editor. All editors on the market output audio along with the video. I think most of us have NOT recorded the binaural audio, we have simply added audio to elements or places in the scene, and the binaural is created via the camera when we play back. So now we have to get some form of third party way of capturing that and then synching, because it can’t be done in Unreal.
If you wish to playback binaurally recorded audio inside UE4, then your best bet is to actually NOT spatialize. This is because sound recorded with a binaural microphone, like the above from 3dio, already has encoded spatialization. So you would play it back non-spatialized.
Binaural audio can be very effective, as you can see here from the Hellblade sound documentary:
Binaural sound is 2 channels (like most stereo sounds), but not all stereo sounds are binaural sounds. Just like a square is a rectangle but not all rectangles are squares.
Binaural recordings work best if the user is listening on headphones, but in the case of Hellblade, liberties can be made for the effect.
You can simulate binaural audio using a processing technique called HRTF:
Plugins like Oculus Audio, Google Resonance, and Steam Audio allow you to take a 3D mono sound source inside UE4 and encode binaural spatial information so that it outputs a stereo sound that feels more spatially resolved.
**However, you’re talking about rendering a video, right? **
This is where my misunderstanding was.
If I were to speculate why Sequencer doesn’t render out to audio, my guess would be that it wasn’t really possible before. The old Audio Engine just sent sources to the device API and sources were mixed by the platform itself. So in the old Audio Engine UE4 never had a mix-down that it could spit out as an asset.
However, it would be possible in the New Audio Engine, so I imagine when the Sequencer team can sit down and look at their rendering pipeline with the New Audio Engine, they’ll be able to develop something along those lines.
Just a guess, but I’m sure eventually we’d like to get that kind of functionality.
The Audio Engine dev team is looking at ways we can render mix assets (so it’s on our road map), perhaps when we get there, the Sequencer team can take the hand-off. That’d be great!
Thanks Dan, for the clarification. Yes I am just talking about stereo. I still would find the ability to output from the sequencer an audio file something that would be very usable
Mike
Yeah, the sequencer team is working on a way to export video and audio from sequencer. We’ve got the ability to record audio from any submix in UE4 now, it’s a matter of recording/exporting audio and putting it in the video file.
Your point of “All editors on the market output audio along with the video” is not exactly true.
UE4 is (and has always been) a game editor for games… I assume the other editors you’re talking about are editors for videos or are DAWs – both applications which are fundamentally designed from the ground up for media asset exports and non-real-time rendering.
It’s a totally new thing to record video (and audio) and render the output from a game engine directly to a single output product, basically turning a game engine/editor into a video production suite.
Audio rendering in a video game engine is very much not designed for non-real-time audio recording/bouncing.
However, we do have a plan to implement a non-real-time recording mode for the new audio mixer in UE4 (so you can do things like increase the # of sources rendered, use ultra-expensive effects, etc), but that’s down the line.
Hello! Knowing it’s an old post, what’s the status of this? It’s 2024 and to record audio from in game to a sequencer I route audio back to unreal with Voicemeeter Banana and use the microphone take recorder functionality. However it’s not the best because it has delay and stuff. Hope it’s still down the line somewhere in dev!
