New NVIDIA Broadcast 1.2 Update Removes Room Echo, Pet Noise And Video Noise To Transform Any Room Into A Home Studio

By Gerardo Delgado on May 11, 2021

The NVIDIA Broadcast app turns your room into a home studio, and today, it’s getting a big update. Version 1.2 includes new effects that enhance your microphone and webcam quality, plus features requested from the streaming community.

NVIDIA Broadcast technologies are also being integrated natively into top streaming apps such as OBS Studio, AVerMedia and Notch.

NVIDIA Broadcast uses AI to improve the quality of your microphone, speakers and webcam. RTX GPU owners can apply AI effects to their own microphone, or any incoming audio, removing unwanted noise. AI features also enhance the camera, including Background Removal, which removes a cluttered background from the camera feed, or Auto Frame which keeps the subject centered in the camera, even if they move. These effects run in real-time using Tensor Cores found on GeForce RTX GPUs. It’s the perfect tool to enhance both live streaming and video conferencing.

 

NVIDIA Broadcast 1.2

As millions of people have pivoted to working from home, one noticeable struggle has been room echo — or reverb. It turns out that bedrooms, living rooms and kitchens — the new home offices for many aren’t ideal spaces for capturing digital audio.

The new Room Echo Removal feature takes care of that by removing the echoey sound of your voice in rooms with poor acoustics. That echo may not feel so bad at first, but after a long streaming session or more than 400 days of meetings from home, your audience or colleagues will thank you for getting rid of it.

In addition to the audio-based Room Echo Removal, bothersome video noise is also going to become a thing of the past. Lower quality cameras often can output “noise” — static that the audience can see on the video feed, especially in low-light situations. This is a complex challenge, one that will take some time to perfect — but the Video Noise Removal beta feature is a first step, helping to reduce video static to deliver a cleaner image.

Existing features also receive improvements. The new update further enhances the quality and performance of the popular audio noise removal tool, with added profiles to better separate the sound of cats, dogs and even insects (cicadas be gone!). Auto Frame is also getting an update with a “buffer zone” that will let the subject move within the frame, without the camera updating. Now, it’ll move only when the talent leaves the middle-third of the screen.

In addition to feedback we receive on the AI effects, we receive a ton of feedback on the app itself. One bit of feedback, now incorporated in the app, is being able to run multiple AI effects on a single device. Now users can combine Background Blur with Auto Frame, or any combination they choose.  

NVIDIA Broadcast Effects in AVerMedia, OBS and Notch

NVIDIA partners closely with the content creator ecosystem, accelerating and enhancing top apps for RTX GPUs.  

Last October, the NVIDIA Maxine SDKs that power the NVIDIA Broadcast app were released so developers could add these effects natively in their apps. Last month, we announced that OBS Studio was integrating our Noise Removal effect natively in OBS Studio 27, and that Notch had also integrated Body Tracking and Virtual Background.

Today, we are announcing that AVerMedia has integrated the audio noise removal and virtual background effects natively into their apps. Users with an AVerMedia microphone or webcam can get these effects directly applied to your device, making it easier than ever to stream perfect audio and video.

 

Stay in Touch!

We hope that you enjoy the updates. Try them out, and please send us your feedback on the NVIDIA Broadcast forums. Browse other comments and suggestions and vote for the ones you like as this will help guide future app development.  You can help us further refine the AI by sharing a short audio and/or video clip at our feedback site. This will allow us to anonymously train the AI on your setup, so future iterations will run even better.

Comments