I’ve been building https://beatstorapon.com
as a vertical, music-only alternative to generic freelance / gig marketplaces (Fiverr, Upwork etc.).
Right now, if you’re a rapper, producer, engineer, playlist owner or promoter, you get thrown into the same bucket as “I’ll design your logo for $5” and “I’ll do data entry”. The category is “Music & Audio”, the incentives and UX are not built around how music actually works.
What BeatsToRapOn is today
A culture-focused platform for music creators and music-adjacent services only
AI tools that already get daily use:
AI stem splitter / vocal remover (Demucs-based, Go + FFmpeg + Python)
Valkyrie AI Mastering tuned for rap/trap/hip-hop/R&B (Python, FastAPI, GPU mastering chain)
Song Key & BPM Finder (Essentia/Librosa)
Simple AI reel generator for social clips
A new music services marketplace where people sell:
Music promotion (playlist pitching, campaigns, etc.)
Mixing / mastering / vocal tuning
Video + reel editing
Cover art / branding / EPKs
Ad inventory / sponsorships to a music-only audience
Queue / jobs: long-running audio jobs via worker processes; experimenting with process isolation vs. throughput
Data: MongoDB for users/listings/events, S3 for assets
Frontend: mostly hand-rolled HTML/CSS/JS (no big SPA framework yet), Wavesurfer.js for waveforms
Why I’m building this
Generic platforms treat music as a side category; the algorithm doesn’t know/care about genre, scenes or long-term careers.
Fees and incentives are optimized for “as many gigs as possible”, not “move this record and grow this catalog”.
A lot of artists already use the AI tools on the site. It felt natural to let them sell what they’re already doing (reels, mixes, masters, promo) inside the same ecosystem instead of sending people off-platform.
Where it’s at
Live tools with real usage (stem splitting, mastering, key/BPM, reels)
Marketplace is live but still early – I’m onboarding sellers and tightening flows/fees before pushing it hard
Revenue right now comes from the AI tools + early marketplace activity
What I’d love feedback on
From a dev/product lens:
Does the architecture choice (Python mastering chain + Go/FFmpeg workers) make sense, or would you simplify it?
Got opinions on handling long-running audio jobs cleanly (FastAPI + workers) without over-engineering?
From a marketplace/vertical SaaS lens:
Is “music-only Fiverr + integrated tools” actually compelling enough?
Where would you simplify or narrow even more (e.g., just focus on promotion / mixing / mastering first)?
I’m very open to blunt feedback. If you think this is the wrong wedge, the wrong UX, or the wrong business model, I’d rather hear it now.
Chet-Fitzgerald•2mo ago
Right now, if you’re a rapper, producer, engineer, playlist owner or promoter, you get thrown into the same bucket as “I’ll design your logo for $5” and “I’ll do data entry”. The category is “Music & Audio”, the incentives and UX are not built around how music actually works.
What BeatsToRapOn is today
A culture-focused platform for music creators and music-adjacent services only
AI tools that already get daily use:
AI stem splitter / vocal remover (Demucs-based, Go + FFmpeg + Python)
Valkyrie AI Mastering tuned for rap/trap/hip-hop/R&B (Python, FastAPI, GPU mastering chain)
Song Key & BPM Finder (Essentia/Librosa)
Simple AI reel generator for social clips
A new music services marketplace where people sell:
Music promotion (playlist pitching, campaigns, etc.)
Mixing / mastering / vocal tuning
Video + reel editing
Cover art / branding / EPKs
Ad inventory / sponsorships to a music-only audience
Marketplace entry point for sellers: https://beatstorapon.com/sell-music-services
Tech stack (for the HN crowd)
Backend: FastAPI (Python) + some Go workers for Demucs processing
Audio: FFmpeg, Librosa, Essentia, Torchaudio (with GPU when available)
Mastering: custom multi-stage Python chain (LUFS normalization, multiband compression, sidechain ducking for 808/kick, etc.)
Queue / jobs: long-running audio jobs via worker processes; experimenting with process isolation vs. throughput
Data: MongoDB for users/listings/events, S3 for assets
Frontend: mostly hand-rolled HTML/CSS/JS (no big SPA framework yet), Wavesurfer.js for waveforms
Why I’m building this
Generic platforms treat music as a side category; the algorithm doesn’t know/care about genre, scenes or long-term careers.
Fees and incentives are optimized for “as many gigs as possible”, not “move this record and grow this catalog”.
A lot of artists already use the AI tools on the site. It felt natural to let them sell what they’re already doing (reels, mixes, masters, promo) inside the same ecosystem instead of sending people off-platform.
Where it’s at
Live tools with real usage (stem splitting, mastering, key/BPM, reels)
Marketplace is live but still early – I’m onboarding sellers and tightening flows/fees before pushing it hard
Revenue right now comes from the AI tools + early marketplace activity
What I’d love feedback on
From a dev/product lens:
Does the architecture choice (Python mastering chain + Go/FFmpeg workers) make sense, or would you simplify it?
Got opinions on handling long-running audio jobs cleanly (FastAPI + workers) without over-engineering?
From a marketplace/vertical SaaS lens:
Is “music-only Fiverr + integrated tools” actually compelling enough?
Where would you simplify or narrow even more (e.g., just focus on promotion / mixing / mastering first)?
I’m very open to blunt feedback. If you think this is the wrong wedge, the wrong UX, or the wrong business model, I’d rather hear it now.
Links again: Main site: https://beatstorapon.com
Seller hub: https://beatstorapon.com/sell-music-services