Terminals and Tunes
Music for coding, focus, and the quiet hours between builds.
Hey folks!
YEN has always had two halves in the name. The terminal half is the part you can see — the app, the IDE, the splits, the tabs, the file browser. The musical half has been quieter and never mentioned.
Well, until now.
I just shipped Terminals and Tunes — an AI-assisted music label that produces background music for the kind of person who reads my blog. Late-night coders, deep-focus readers, anxiety-relief sleepers, meditation-curious focus types, cottagecore journal-keepers. '
Five personas. Same channel. Same person behind it. Different worlds.
This post is about why I built it, what’s live right now, and how the pipeline actually works.
Why Music?
The terminal-first IDE thesis I’ve been writing about for two years is “the developer surface should be the terminal.” Background music is the same kind of surface. It’s the thing playing in the corner while you work — quiet, dependable, doesn’t need attention, doesn’t sell you something every thirty seconds.
The Lofi Girl channel earns reportedly six figures a month doing exactly this, with one persona and a finite catalog. The category is real. The audience is real. The competitive moat is taste plus volume plus distribution discipline — all things I can build with CLI tools.
So I built one to see if it would really work.
It kind of does.
The 5 Personas
I picked five niches I think pair well with the YEN: Terminals and Tunes brand and don’t overlap with the giants:
YEN Terminal Tunes — synthwave for late-night coding. Restrained, melodic, low-distraction. Background music for the terminal.
Idle Coast — bossa nova / lounge / chillhop. For cafes, slow afternoons, and golden-hour workspaces.
Noise Floor — brown noise plus soft ambient pads. For sleep and deep relaxation.
Gamma Daemon — binaural focus / 40Hz gamma. For study sessions, deep focus, and meditation.
Hearth 16 — cottagecore acoustic folk plus nature ambience. For reading, journaling, and quiet evenings.
Each persona has its own anime-themed visual identity, its own audio fingerprint, its own niche audience. They all roll up under one YouTube channel (@yenFTW) and one Spotify For Artists account, so I claim once and pitch all five.
The Pipeline
Every step is CLI or scripted. My role is taste and quality control. The actual work is:
Generate — Suno Premier through aimusicapi.ai. I send a prompt; the API returns audio. About $0.10 per track.
QC — A TUI I built. I press k to keep, r to reject, p to preview. Hard taste gate. Most tracks get rejected.
Mix — ffmpeg crossfade for long-form mixes. Loudness-normalize to -14 LUFS for Spotify.
Cover — Higgsfield’s kling3_0 at 4K renders one ten-second seamless loop per persona. Every release cover is a frame extracted from that loop, so the album art always matches the YouTube video’s character, world, and lighting.
Video — ffmpeg layers the seamless loop under the audio at 4K. Audio is 320 kbps AAC, video is H.264 high-profile at CRF 16.
Metadata — claude -p writes the YouTube title, description with chapter markers, and 15-25 SEO-tuned tags.
Hard upload gate — Operator approval required before anything publishes. I review every file. Nothing gets posted on auto-pilot.
Upload — YouTube Data API for the video (with containsSyntheticMedia: true because the AI disclosure is mandatory and honest), Supabase Storage for the durable archive, DistroKid for streaming distribution.
The whole thing runs from a slash command: /yen-tunes generates per-persona in parallel, then converges at QC, then fans back out for per-mix cover and video. End-to-end it’s about ninety minutes of operator time and $1-2 in API credits for a single pass across all five personas.
Not too bad.
It’s just an experiment and I’m still messing with the settings and outputs as well as the distribution to other platforms. It’s not perfect but it’s amazing what you can do with a little knowledge (and an agentic harness).
I can’t imagine a world where music isn’t part of my programming “logic” and so I’ve had a blast experimenting and seeing how it all fits together.
I hope you enjoy.
— 8
If you want to listen, the personas and playlists are here: