When NVIDIA introduced the free NVIDIA broadcast app in 2020, they promised to use the AI features of the RTX GPU to improve the video and sound quality of users’ recordings, assuming that they have the right hardware. This is useful when the laptop’s built-in webcam and microphone are unable to capture the best-looking footage and richest audio, and it could help streamers use cheaper setups. Version 2.0 of the NVIDIA Broadcast App released at the end of January makes the feature even higher… perhaps a little too close to the sun.
With NVIDIA Broadcast 2.0, microphones still have noise removal and room echo removal options, but beta also has a studio audio effect that “improves the quality of the microphone to simulate a high-end recording studio.” For videos, the app continues to adjust to the background (swap, blur, or essentially green screening), remove noise from particle-rich footage, zoom in and hold it automatically in frames, and weird things To make it look like you’re looking at the camera. However, the new feature in the update is virtual lighting. You can showcase you better with videos.
Some of these new features are looking for a powerful GPU. Studio audio, video noise removal, and virtual keylights are all looking for “high-end GPUs” and are not recommended when using games or other GPU-intensive applications. All features require RTX hardware. So you’ll need at least an RTX 2060 or higher to try them out, but for features that require a “high-end GPU”, Nvidia says “need” the RTX 4080, 5080 or higher. ” However, I was able to perform both functions on my RTX 4060 mobile GPU.
Nvidia will be broadcast during the event
The ideas behind these AI features are cool, but how well they actually work is still a question. For one thing, they may be really demanding, as Nvidia says. Running virtual keylight or studio audio functions on a laptop with an RTX 4060 has shown that the GPU is absolutely denounced by the process. Nvidia’s embedded GPU usage monitor was displayed in red. The RTX4060 was all at its peak, and the performance overlay attracted 60 watts. My laptop fan started up as if I was playing a game at full throttle. So, from an economics perspective, these features are costly no matter how you look at them. They require strong hardware to run, and you need to run that hardware strongly. Plan to connect your desktop computer or laptop to use these features.
Then there’s even more important questions about how they look and what the sound looks like. Let’s start with the video.
Credit: Marknap
Despite being available before Broadcast 2.0, the eye contact tool has come out of beta. But I’m not sure it should have. Certainly, making that possible makes it look like you’re staring at the camera in video footage. But in my tests it always gave me big blue eyes. For reference, I don’t have blue eyes. Even when I was making eye contact with the camera, Nvidia broadcasts still insisted on editing the eyes and making them blue.
Credit: Marknap
Virtual Keylight did what it said. I created artificial lighting to enhance the brightness for me without increasing the brightness of the entire video. But the outcome did not impress me. When enabled, it just looks radioactive. The lighting is very unnatural.
Audio sounds pretty impressive at first. My laptop’s microphone is not very good. Even in the quiet room they gave me a sound that echoed in the distance and slightly muffled. With Studio Voice enabled, my voice is much more fulfilling and sounding more clearly. But, to hear it, there’s a strange digitalization going on. It’s hard to characterize, but it really doesn’t sound like my voice. It sounds like my voice recording was being used with audio cronar, but it repeated everything I said. It’s all a bit sturdy and trembling. Listen to:
The Studio Voice feature can’t save the microphone from a bad recording environment either. In tests in a small room where the box fans were running in full blast, the audio dramatically improved clarity compared to the raw recordings from the microphone, but still tuned. It’s particularly strange.
If you have half the microphone, the studio’s voice may make it worse. I created additional test recordings wired directly to my laptop using a built-in boom microphone in my Audeze Maxwell headset. Both in the quiet and large rooms, provided a loud, clear and complete recording of my voice without enabling the studio voice. In both cases, turning on studio audio introduced digitization that makes mistakes difficult, which not only made the audio sound worse, but also made it difficult to understand.
Can nvidia ai replace the right streaming setup?
Given the hardware requirements, performance requirements, and the quality of the results, the stars need to tweak these new Nvidia broadcast features to feel truly valuable. If you have a system with NVIDIA, always use the tools to play. Some features can be useful, like autoframing. However, we don’t recommend the new Nvidia GPU schent, especially if you want to get closer to professional quality, so you can save money with audio and video recording gear. Also, don’t forget that the power draw of the GPU that attempts to perform these functions will be summed over time.
Audio quality obtained from Studio Voice – probably limited by my system’s RTX 4060, but it paled in comparison to the quality that I can get, not something I want to share with every audience on a regular basis. Because I have a headset with a boom microphone. I’ve tested many gaming headsets, and even a very inexpensive wired headset with a boom mic is better in the league than what I’ve heard in the studio’s voice.
The eye contact function was nothing more than anxious. I don’t think anyone will be fooled by believing that they are actually making eye contact. And virtual key lights that resemble the voices of the studio seem to be a quality alternative to actual key lights, especially when affordable LED lights are a dozen.
