BTS LIVE AUDIO vs. BROADCAST AUDIO // How AI can affect what audiences hear at home LISTEN TO THE DIFFERENCE: Clips 1 and 2 are the same “MIC Drop” performance: the likely AI-suppressed broadcast audio from Disney+ and Hulu vs. BigHit’s human audio engineer and sound captured by their... drones. Clips 3 and 4 are the same fan cam: first with its original sound, then after AI was used to enhance the voices and suppress the crowd. This is a ROUGH example to show how AI can accidentally get confused. Remember when we pointed out that parts of the broadcast sounded like one ARMY in the audience was singing with BTS and you could hear it? Also, that the mics were not equal and went unnaturally low? That’s a classic, telltale effect of aggressive AI sound suppression. Watch the end of this video example. Listen for: 1) The ARMY singing along with Tae (AI picked up the audience) 2) the unnatural high-low (Jin and Tae are low, and then it JUMPS for Yoongi, Jungkook and you can hear Hobi) This is what auto-AI does. WHY DOES AI GET CONFUSED? Voices, instruments and crowd noise can occupy the same frequencies. AI has to reverse-engineer the audio and guess which sound belongs where—and the technology is not perfect. HOW DOES THAT CHANGE WHAT WE HEAR? While suppressing the crowd, AI can accidentally cut frequencies from the singers’ microphones, causing sudden volume drops and unnatural low or high pitches. CAPCUT TEST: You can test this yourself using CapCut’s AI Voice Enhance and Crowd Fade. The features are free to test. Try it and see! It’s always fun to experiment :) FOR REFERENCE: L-Acoustics’ Source Intelligence is one example of live AI software that major broadcasters use for TV that isolates vocals from crowd noise and instrumental bleed. I’m NOT saying Disney+ or Hulu used that specific program. Rather, pointing out it does exist for major broadcast and is used. BTS AT IHEART RADIO MUSIC FESTIVAL 2026 BTS BTS BTS BTS PAVED THE WAY #BTSxIHeartRadioshow more

Fan Cam used in example:

Do you have confirmation from somewhere that iheart was using AI during the live broadcast? Is that an industry standard? Genuinely asking cause I don’t know how these festivals work.

Good and fair question! No, I never confirmed that. I only pointed out that the broadcast audio had several telltale signs of AI sound suppression. The sound from the broadcast was unnaturally low. If you follow the quoted retweets, the original (third going backwards) tweet is a live recording from my iPhone while I was recording the broadcast. One telltale sign is when the audio suddenly goes unnaturally high and low or sounds uneven. When the vocals, instruments and crowd overlap, AI can get confused about which sound belongs where. It blends them all together then separates it. Another big telltale is when it accidentally brings individual voices from the audience to the front. If you listen to the live broadcast during MicDrop and FYA — it sounds in some parts like one single ARMY is singing with BTS. That also happened in the example I made at the end of the video (4th video clip.) In my example at the end, AI did exactly that. I didn’t even have to try to replicate what we heard on Disney+ and Hulu—it happened naturally because the technology simply isn’t there yet. Not all broadcasters use AI. Traditionally, the singers, backup singers, instruments and crowd microphones all have their own audio feeds. A human audio engineer balances them in real time, knowing what to raise, lower, mute or unmute and how much crowd sound to fade in or out. Depending on the show, a live broadcast may run on a 5–20-second delay, but the audio is still mixed live. Dynamic microphones also help. Dynamic microphones are built to pick up the sound directly in front of them while using opposing sound waves to reduce sound coming from elsewhere, including the audience. Basically, one sound wave helps cancel another—similar to noise-canceling headphones. Audio engineers can also use this type of sound cancellation and other controls to reduce the audience noise coming through the microphone feeds. Human audio engineers have every feed listed separately on their consoles. They can hear each layer, know what to listen for and adjust it before everything is combined into the final broadcast mix—just like music recorded in a studio. It’s layers. Think like creating something in Adobe Illustrator or Photoshop — it’s in layers, but audio. Does that make sense? So while never confirmed that they used AI — there were 2 major things that happened in the Disney+ and Hulu broadcast that are big signals that they did use AI and not a human audio engineer.