You’re invited into a legacy family audio business that refused to accept “good enough” on feedback control and instead chased the impossible: a truly zero‑latency, AI‑driven way to push your PA louder without squeals. You follow Devin Sheets from growing up on sound gigs to roaming European stages, then back home to build De‑Feedback plugin for working musicians, a live sound feedback plugin and on‑the‑fly impulse‑response generator that listens like a seasoned engineer: separating human voice, room reverb, background noise, and feedback in real time so you can grab at least 6 dB more gain before things start to howl. Along the way you see how NAMM sparked the idea, how inverse impulse responses and probability math beat old EQ and gate tricks, and how “homebrew AI” meant sneaking into every empty church at 3 a.m. just to teach the model what real rooms actually sound like.
You also learn how to think like a modern working musician: using social media to find the right AI programmers across the world, leaning on LLMs to translate, collaborate, and even rate contractor work so you can move faster without losing control. You come away knowing you can drop a dedicated De‑Feedback box or plugin into almost any rig, from churches to touring consoles to tiny clubs, take it with you even when someone else is behind the board, and quietly stack the deck in your favor. In the end, it’s a roadmap for how you run your own gigs and career: stay curious, embrace new tools, protect your sound, and Always Be Performing.
- 00:00:00 Gig Gab 524 – Monday, March 9th, 2026
- March 9th: National Meatball Day
- Guest co-host: Devin Sheets from Alpha Labs
- 00:02:12 Let’s Grow this Legacy Family Business
- Grew up doing sound
- Also a musician
- Lived in Europe
- Then came back and said, “let’s grow this family business!”
- 00:03:44 We haven’t “just solved” this feedback problem
- Went to NAMM for the first time, and was inspired
- There are automated EQ-based or gate-based systems
- PSE plugin from Waves
- 5045 for feedback
- 00:04:57 Why isn’t there a “balanced audio”-type solution for Feedback
- Balanced Audio fixes hums and it just works.
- 00:08:24 NAMM is a great inspiration…and it inspired Devin and his team to seek a feedback plugin solution
- People get entrenched
- Inverse Impulse Response methodology
- 00:12:35 Training the AI to listen for three things: human voice, reverb, and feedback
- Created a de-reverb algorithm and went beyond that
- A probability calculation does the math
- 00:16:05 Truly zero latency for the plugin
- Workflow latency remains
- 00:19:32 I don’t have any coding or AI background, but I have a gut feeling AI will fix this feedback problem
- Others: It’s harder than you think
- Devin: I knew that it needed to happen
- 00:20:58 Finding an AI programmer who was interested in doing
- Experimented with some programmers, failed, learned some things!
- 00:21:09 Social Media to the rescue!
- Late 2023: Devin found a group of AI programmers who would be interested
- Sending large amounts of money to China…it’s a risk!
- 00:26:30 At 3am, a text message: I think I’ve done it.
- Devin immediately started testing it himself
- “It seemed to work.”
- 00:27:17 Installing De-Feedback in Churches
- Sponsors
- 00:30:57 SPONSOR: Claude.ai – Ready to tackle bigger problems? Sign up for Claude today, which includes access to Claude Cowork, too, when you visit Claude.ai/giggab
- 00:32:43 SPONSOR: Squarespace. Check out https://www.squarespace.com/GIGGAB to save 10% off your first purchase of a website or domain using code GIGGAB.
- 00:34:20 What is an impulse response?
- Impulse Response: An audio picture of how the room sounds
- Popping balloons in a room/environment and recording the sound is a common approach for creating impulse responses
- 00:38:33 De-Feedback is an on-the-fly IR generator
- …and analyzer that’s trained on the human voice, room reverb, background noise…and feedback
- 00:41:55 Finding the right programmers was the key
- …in addition to actually having the idea and the bullheaded persistence to make it happen.
- 00:44:46 Mind-melding was necessary
- And LLMs helped with translation!
- 00:48:39 Using AI to make it possible to collaborate with other humans
- 00:50:03 Using an LLM to rate the work of your contractors and employees
- 00:51:54 How do we get De-Feedback into the hands of working musicians
- US$499 for the De-Feedback plugin
- VST3 or AU plugin
- A higher-end Windows laptop can likely run it on its own
- Apple’s Core Audio tech makes it difficult, but they’re working on it.
- De-Feedback also sells a perfectly-tuned headless computer to do this
- Alpha Labs tried tons of interfaces that the Focusrite Scarlett keeps glitches out of the mix
- Waves SuperRack LiveBox
- 01:01:37 Where do we expand?
- 01:05:18 Homebrew AI!
- Training EVERY room he could find
- “Can you let me into your empty church at 3am?” – To record IR to then train the data set for De-Feeback
- 01:07:25 Creating your own AI model
- 01:08:13 What’s the future look like?
- Acquisition? Demands for security? – Planning for it all
- 01:09:26 You can get this and bring it with you to gigs where someone else is doing sound
- De-Feedback Option 1
- Allen & Heath Qu-5’s Feedback Eliminator
- De-Feedback gets at least as 6dB more gain before feedback
- 01:17:46 Gig Gab 524 Outtro
- Follow Devin Sheets
- Contact Gig Gab!
Podcast: Play in new window | Download (Duration: 1:20:15 — 73.9MB) | Embed
Subscribe: Apple Podcasts | Spotify | Podcast Index | RSS | More
0 Comments