AI Music x Real Footage: Creating 'Drifted: Letters from Jeju' – A Lo-fi / Neo-Soul / Bossa Nova Album with ChatGPT, Suno & iPhone Filmmaking

:headphone: Project Overview

Drifted: Letters from Jeju | 제주에서 띄운 편지 is a lo-fi, neo-soul & bossa nova concept album that blends ambient AI-generated music with real footage filmed along the breezy coastlines of Jeju Island, South Korea.

Created as an experiment in human–AI collaboration, it brings together:

  • ChatGPT-4o – for concept, lyric writing (Korean tracks), and Suno prompt development

  • Suno AI v4.5 – for all audio/vocal composition

  • iPhone 16 Pro Max + DJI Osmo Mobile + DJI Mimo – used for on-site filming.

  • Wondershare Filmora – for streamlined video editing

:movie_camera: Footage was shot entirely on location. The DJI Osmo Mobile gimbal and DJI Mimo app’s smart features enabled smooth, stable, high-res iPhone recording—even in Jeju’s shifting coastal winds—with no post-stabilization required.

The goal was to match real-world light, motion, and emotional tone with AI-assisted musical storytelling—grounded, not synthetic.


:musical_score: Highlighting Track 3 – By the Sea, Just Us

One highlight of the album is Track 3: “By the Sea, Just Us” | 바닷가엔 오직 너와 나.

It was conceptualized with ChatGPT-4o, using this exact Suno style prompt:

A soft, emotional Korean-language lo-fi soul ballad with warm electric piano chords, mellow lo-fi drums, gentle acoustic guitar textures, and subtle gayageum accents. Female vocals delivered with an intimate, wistful tone—sincere and heartfelt, like a love letter whispered beside the sea. The lyrics are poetic and reflective, capturing a timeless romantic moment between two people on Jeju Island. The atmosphere should feel dreamy, nostalgic, and calm, with subtle vinyl warmth and ambient sea breeze textures. Perfect for sunset listening or quiet emotional reflection.

The resulting track blends poetic Korean lyrics (also co-written with GPT) and nostalgic instrumentation, evoking timeless love and serenity on Jeju’s shore.

:headphone: [Listen to Track 3 directly: 06:04 Timestamp]
:play_button: https://youtu.be/HT72Z2Zj_2w?t=364


:robot: Behind the Collaboration

A quick breakdown of the AI-driven creative process:

ChatGPT-4o was used for:

  • Developing the narrative arc of the album
  • Writing the Korean lyrics for Tracks 3 & 8
  • Iterating on musical style prompts grounded in place and mood

Suno v4.5 (Suno model) was used for:

  • Composing all vocals and instrumentation
  • Shaping genre-driven tracks with emotional texture

:speech_balloon: What I’d Love to Discuss with You:

  1. Prompting Suno for Mood + Emotion
    → Any tips on shaping phrasing, voice tone, or dynamic arc beyond basic genre?

  2. GPT for Musical Ideation
    → I used GPT to shape lo-fi/neo-soul/Bossa prompts that felt grounded in Jeju. How do you develop style input prompts that feel cinematic or place-based?

  3. Low-Cost Production Pipelines
    → Have you used similar mobile filming + AI audio workflows? Any tips or toolchains to share?

  4. Music-Video Sync
    → I’d love to hear how others handle aligning ambient AI tracks with real-world visual pacing and scene dynamics.


If it’s helpful, I’m happy to share:

  • Exact Suno prompts for each track
  • How I matched lyrics and footage
  • Editing strategies for musical flow and narrative rhythm

Looking forward to learning how others are blending the real and generative in audio/visual projects!

SynthEchos