We have already dealt with RTX Voice several times, now there is Broadcast, where NVIDIA can process the image capture of the camera on the one hand and also includes RTX Voice on the other hand. Unfortunately, the whole thing is RTX-exclusive so far, so no bending of the installation package will help at the moment. But be that as it may, it’s well worth it, otherwise I certainly wouldn’t have put in the effort as well. We remember the microphone and background noise from RTX Voice. This works excellently with the recording device and can even be recorded and checked with a built-in test program. You shouldn’t overdo it with the strength or it will clip a bit. But otherwise the whole thing is certainly really helpful when the vacuum cleaner, coffee machine or hair dryer are making noise in the background. This also works during playback, if the other person doesn’t have such a feature, then you can have this noise carpet calculated again before playback, thanks to AI.
But now comes the real highlight, which is the AI-based recognition of the protagonist in front of the cam and manipulating what is not me. I didn’t just invest a thousand euros in my green-wall studio and then this comes along for free. Well, the edges are a little frayed and with fast movements there are partial also sometimes slight artifacts (Finder), but these are mostly provoked exceptions or nasty backlight.
It also works as a foreign studio, such as here as a correspondent in Taipei. If you can’t fly there, you just put a can in the background and turn up the heat. Then even the forehead looks authentic. You have to give Broadcast credit for the fact that a lot stands and falls with the quality of the camera. Unfortunately, the one in the Razer Blade is rather poor and faint, so many artifacts already existed even without an additional program. But what works well is the realtime stream, which is much more performant with NvEnc than with the CPU. Nice side effect: the notebook’s fan stays discreetly in the background because the GPU does it much more efficiently.
Overall, of course, you have your fun, but especially for video conferencing, a more official background is better than the chaos in the home office. Or you can record the whole thing as a video file. MKV container or MP4, again NvEnc helps with real time encoding without the rather pedestrian CPU having to struggle one off. A test with two screens running two different apps doesn’t bring the notebook to a standstill, because the load is almost solely on the GPU.
As with RTX Voice, all of this can be easily selected as a recording or output device in OBS. This can be done in any software that allows device selection. Skype, Teams & Co do it too, we’ve already seen OBS. Here you can just pick and set up the device, whether it’s video or audio. Complicated is definitely something else.
In addition, more and more companies are also using NVIDIA’s broadcast API and features to implement directly in apps or plugins as well. For example, VoiceFX brings NVIDIA Broadcasts noise removal to VST3-compatible editing apps. Xaymar, the developer of the popular StreamFX plugin for OBS Studio that already uses the AR SDK, has now added noise removal to a VST3 filter with VoiceFX. This enables all users to take advantage of NVIDIA’s Noise Removal feature to remove background noise directly from video and audio editing applications such as Adobe Premiere Pro and Adobe Audition. Even after the fact.
21 Antworten
Kommentar
Lade neue Kommentare
1
Mitglied
Veteran
Urgestein
Urgestein
Urgestein
Urgestein
Urgestein
Mitglied
Urgestein
Urgestein
Urgestein
Urgestein
Urgestein
Mitglied
Urgestein
1
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →