AI-native audio plugin host control via Model Context Protocol. Enables Claude and other LLMs to interact with professional audio plugins through Carla.
TypeScript MCP Audio AI
Andrew Grathwohl
Software Developer, Systems Architect & Audio Engineer
Nashville, Tennessee
About
Software developer and audio engineer. I build real-time DSP systems, MCP servers for audio applications, and P2P infrastructure. US Patent holder for algorithmic voice casting.
I've been writing DSP algorithms, deploying P2P infrastructure, and building audio tooling since 2011. Started at Audible automating QC. Ended up with a patent and a distributed transcoding network. Now in Nashville running Tube.sh and building MCP servers for audio production.
Technical Expertise
Languages
RustCC++SwiftNode.jsTypeScriptPythonGolangElixir
Frameworks
Next.jsReactElectronTauriVue 3RemixReact NativeFastAPIPhoenix
Audio DSP / ML
PyTorchDemucslibrosapedalboardJACKFFmpegSoXrubatorustffthoundpyebur128pyloudnormTorchAudioSpeechBrainSpaCy
Databases & Storage
PostgreSQLSupabaseHAMMER2GlusterFSZFS/OpenZFSBtrfsCephIPFSJuiceFSXFS
Infrastructure
GCPAWSDockerKubernetesTerraformNixAnsibleModalVercel
Networking
bird2FRRNginxVarnishNetboxntop-ngKamailioFreeSWITCHWebRTCSTUN/TURN
System Tools
sysctlcpupowerbcc/bpftraceperftcpdumpRT_PREEMPT
Operating Systems
GNU/LinuxFreeBSDDragonFly BSDmacOSVyOSNuttXpfSenseRouterOSJunos OSNixOS
Open Source Projects
Real-time text protocol converter for LLM streaming. Converts LLM output to ITU-T T.140 format for accessibility and telecommunications.
TypeScript Accessibility LLM Real-time
MCP server enabling AI agents to interact with SuperCollider for algorithmic composition and sound synthesis.
TypeScript MCP SuperCollider Music
Accessibility service for LLM streaming output. Makes AI responses accessible through screen readers and assistive technologies.
TypeScript Accessibility LLM
Uses LLM text generation to compose music algorithmically. Generative audio toolkit.
Music AI Composition
Pair Vibecoding Protocol. Structured approach to collaborative AI-assisted coding.
Development AI Collaboration
Platforms & Products
Tube.sh
Peer-to-peer creator platform for independent media distribution
Part of the Problem
Premium subscription service for Dave Smith's political podcast (150k downloads/episode)
multipli.city
Sonic multiplicities platform for experimental audio
Sonic Multiplicities
YouTube channel for experimental audio and electronic music performances
inbox.audio
Email-to-audio summarization
pvp.codes
Pair Vibecoding Protocol for collaborative AI-assisted coding
mediocre
Uses LLM text generation to compose music algorithmically
Rob Bernstein Comedy
Comedy and podcast production platform
Interviews & Media
Experience
March 2025 - Present
May 2025 - February 2026
October 2024 - April 2025
July 2024 - September 2024
December 2023 - Present
August 2020 - September 2023
August 2017 - June 2020
April 2016 - August 2017
January 2016 - April 2016
September 2015 - December 2015
Backend Engineer
Overture.me
New York, NY
December 2014 - April 2015
September 2013 - December 2014
October 2011 - September 2013
US Patent #9412395 — Narrator selection by comparison to preferred recording features (Algorithmic voice casting using spectral analysis, GMMs, and neural networks)
Education
Bachelor's of Science, Recording Arts
Indiana University, Jacobs School of Music — Computer Music Concentration
August 2007 - July 2011