Replace Spotify API with screenshot taste workflow and example profile.

Remove spotify integration; add TASTE-FROM-SCREENSHOTS guide; ship Ozan settings.json and taste_seeds.json as Cursor examples plus new wandering dervish track. Co-authored-by: Cursor <cursoragent@cursor.com>
2026-06-07 14:47:47 +01:00
parent 6843ecd6b0
commit 02ea026851
21 changed files with 294 additions and 210 deletions
@@ -1,6 +1,6 @@
 ---
 name: ozan-radio
-description: Operate Live Ozan Radio — DeepSeek DJ plans tracks, Google Lyria 3 generates them, Spotify supplies taste. Use when the user wants to start the radio, generate a track, skip, or tune the station.
+description: Operate Live Ozan Radio — DeepSeek DJ, Lyria 3, taste from settings.json and taste_seeds.json. Use for start/generate/skip, or updating taste from Spotify screenshots (no API).
 ---

 # Live Ozan Radio
@@ -8,12 +8,17 @@ description: Operate Live Ozan Radio — DeepSeek DJ plans tracks, Google Lyria
 ## When to use

 - User says "ozan radio", "live radio", "generate a track", "what's playing"
+- User sends **Spotify screenshots** — update `settings.json` + `taste_seeds.json` (see `docs/TASTE-FROM-SCREENSHOTS.md`)
 - Operating or debugging `tinqs/live-radio`

+## Taste from screenshots
+
+**No Spotify API.** Read `docs/TASTE-FROM-SCREENSHOTS.md` for the copy-paste Cursor prompt. Infer genres/mood/instruments from Home, Library, Daily Mixes; write both JSON files; never suggest playing Spotify URLs.
+
 ## Prerequisites

 - `GEMINI_API_KEY` and `DEEPSEEK_API_KEY` in `.env`
- Server running: `python -m ozan_radio serve` (port 8787)
+- Server: `python -m ozan_radio serve` (port 8787)

 ## Operations

@@ -22,18 +27,14 @@ description: Operate Live Ozan Radio — DeepSeek DJ plans tracks, Google Lyria
 | Start station | `python -m ozan_radio serve` |
 | Generate track | `POST http://127.0.0.1:8787/api/generate` |
 | Now playing | `GET http://127.0.0.1:8787/api/now` |
-| Skip | `POST http://127.0.0.1:8787/api/skip` |
+| Skip / shuffle | `POST http://127.0.0.1:8787/api/skip` |
 | One-shot CLI | `python -m ozan_radio generate` |

 ## DJ behavior

-DeepSeek reads Spotify taste (if configured), avoids recent titles, outputs a Lyria prompt. Do not suggest playing Spotify URLs — generation only.
+DeepSeek reads `settings.json` + `taste_seeds.json`, avoids recent titles, outputs Lyria prompts. Generation only — no catalog playback.

 ## Models

 - Lyria: `lyria-3-pro-preview` (songs) or `lyria-3-clip-preview` (30s)
- DeepSeek: `deepseek-chat` via Tinqs inference proxy by default
-
-## Mac live layer
-
-Magenta RealTime 2 (`pip install "magenta-rt[mlx]"`) for real-time beds — see README.
+- DeepSeek: `deepseek-chat` via Tinqs inference or `https://api.deepseek.com/v1`
@@ -8,10 +8,8 @@ DEEPSEEK_BASE_URL=https://api.deepseek.com/v1
 DEEPSEEK_API_KEY=
 DEEPSEEK_MODEL=deepseek-chat

-# Spotify taste profile (optional — DJ works without it)
-SPOTIFY_CLIENT_ID=
-SPOTIFY_CLIENT_SECRET=
-SPOTIFY_REFRESH_TOKEN=
+# Taste: settings.json + taste_seeds.json (see docs/TASTE-FROM-SCREENSHOTS.md)
+# No Spotify API — use Cursor + library screenshots instead.

 # Radio server
 RADIO_HOST=127.0.0.1
@@ -4,13 +4,16 @@ Public repo under `tinqs/live-radio`. AI agents run the station — humans liste

 ## Identity

- **No catalog playback.** Spotify is taste input only. Every track is generated.
- **DeepSeek** plans mood + Lyria prompts. **Google Lyria 3** renders audio.
+- **No catalog playback.** Every track is generated by Lyria 3.
+- **DeepSeek** plans mood + Lyria prompts.
+- **Taste** from `settings.json` + `taste_seeds.json` — typically built from **Spotify screenshots in Cursor** (no Spotify API).
 - Respond in English.

-## Taste
+## Taste setup

-Edit `settings.json` — DJ + chat read it every request. Default: ethnic world dubtronica.
+1. Read **`docs/TASTE-FROM-SCREENSHOTS.md`**
+2. User attaches Spotify screenshots → update `settings.json` and `taste_seeds.json`
+3. DJ reloads both on every generate and chat

 ## Session start

@@ -32,7 +35,6 @@ curl -X POST http://127.0.0.1:8787/api/generate
 |-----|---------|
 | `GEMINI_API_KEY` | Lyria 3 |
 | `DEEPSEEK_API_KEY` | DJ brain |
-| `SPOTIFY_*` | Optional taste |

 ## Siblings

@@ -16,10 +16,14 @@ python -m ozan_radio serve

 Player: `http://127.0.0.1:8787/player`

+## Taste
+
+Screenshot Spotify → Cursor → `settings.json` + `taste_seeds.json`. See `docs/TASTE-FROM-SCREENSHOTS.md`. No Spotify API.
+
 ## Architecture

 ```
-Spotify taste ──► DeepSeek DJ ──► Lyria 3 ──► radio_cache/*.mp3 ──► FastAPI stream
+settings.json + taste_seeds.json ──► DeepSeek DJ ──► Lyria 3 ──► songs/*.mp3 ──► FastAPI stream
 ```

 Optional: Magenta RealTime 2 on Mac for live MIDI/text steering (~200ms latency).
@@ -1,6 +1,6 @@
 # Live Ozan Radio

-Personal AI radio — **no catalog music, ever**. DeepSeek is the DJ. Google **Lyria 3** composes every track. Taste comes from `settings.json` (and optionally Spotify).
+Personal AI radio — **no catalog music, ever**. DeepSeek is the DJ. Google **Lyria 3** composes every track. Taste comes from `settings.json` + `taste_seeds.json` — usually built from **Spotify screenshots in Cursor** (no Spotify API).

 Inspired by [Magenta RealTime 2](https://magenta.withgoogle.com/magenta-realtime-2) (live, ~200ms) and [Lyria 3](https://deepmind.google/models/lyria/) (full songs via Gemini API). On Mac you can layer MRT2 for true live improvisation; this repo ships the cross-platform Lyria + DeepSeek stack first.

@@ -10,8 +10,7 @@ Inspired by [Magenta RealTime 2](https://magenta.withgoogle.com/magenta-realtime
 |-------|---------|------|
 | DJ brain | DeepSeek (Tinqs proxy or BYOK) | Mood, prompts, chat, variety |
 | Music engine | Google Lyria 3 Pro / Clip | Generate MP3 tracks |
-| Taste | `settings.json` + `taste_seeds.json` | Genres, mood, instruments — DJ reads every request |
-| Taste (optional) | Spotify Web API | Top artists, genres — never plays Spotify |
+| Taste | `settings.json` + `taste_seeds.json` | Profile from screenshots or manual edit — see below |
 | Player | FastAPI + `gateway/player.html` | Stream queue, library, chat, dashboard |
 | Live (optional) | Magenta RealTime 2 | Apple Silicon only — see below |

@@ -37,9 +36,15 @@ When shuffle is on (default):

 Daily generation stats live in `songs/stats.json` (gitignored, local runtime only).

+## Taste from Spotify screenshots (recommended)
+
+**No Spotify API keys.** Screenshot your library (Home, Daily Mixes, playlists), open this repo in **Cursor**, attach images, and paste the prompt from **[docs/TASTE-FROM-SCREENSHOTS.md](docs/TASTE-FROM-SCREENSHOTS.md)**. The agent updates `settings.json` and `taste_seeds.json`.
+
+Share that doc with friends — same flow for group taste (WhatsApp links + screenshots).
+
 ## Taste (`settings.json`)

-Edit `settings.json` at the repo root. The DJ reloads it on every generate and chat.
+Edit `settings.json` at the repo root, or let Cursor fill it from screenshots. The DJ reloads it on every generate and chat.

 ```json
 {
@@ -58,7 +63,7 @@ Edit `settings.json` at the repo root. The DJ reloads it on every generate and c
 }
 ```

-Default taste profile: **ethnic world dubtronica** (global roots + dub space + electronic groove).
+**Example profile (Ozan):** checked-in [`settings.json`](settings.json) + [`taste_seeds.json`](taste_seeds.json) — Anadolu psych + ethnic dubtronica from Spotify screenshots. Copy the structure for your own taste.

 The player settings panel PATCHes `playback` and `limits` via `/api/settings` and writes back to this file.

@@ -82,7 +87,8 @@ python -m venv .venv
 .venv\Scripts\activate
 pip install -e .
 copy .env.example .env
-# Fill GEMINI_API_KEY + DEEPSEEK_API_KEY (and Spotify if you have them)
+# Fill GEMINI_API_KEY + DEEPSEEK_API_KEY
+# Taste: docs/TASTE-FROM-SCREENSHOTS.md (Cursor + Spotify screenshots)

 $env:DEEPSEEK_BASE_URL = "https://api.deepseek.com/v1"
 python -m ozan_radio serve
@@ -102,7 +108,6 @@ python -m ozan_radio generate
 | `GEMINI_API_KEY` | Yes | [Google AI Studio](https://aistudio.google.com/apikey) — Lyria 3 |
 | `DEEPSEEK_API_KEY` | Yes | Tinqs proxy token or DeepSeek direct |
 | `DEEPSEEK_BASE_URL` | No | Default `https://api.deepseek.com/v1` |
-| `SPOTIFY_*` | No | Refresh token flow — taste only |
 | `LYRIA_MODEL` | No | `lyria-3-pro-preview` (default) or `lyria-3-clip-preview` |
 | `RADIO_OUTPUT_DIR` | No | Default `./songs` |

@@ -115,15 +120,6 @@ python -m ozan_radio generate

 At the default cap of 10 new songs/day with Lyria Pro, projected max spend is **~$0.82/day**. Adjust `costs` and `limits` in `settings.json` or the player settings panel.

-### Spotify setup (optional taste)
-
-1. Create an app at [Spotify Developer Dashboard](https://developer.spotify.com/dashboard).
-2. Add redirect URI `http://127.0.0.1:8888/callback`.
-3. Complete OAuth once to obtain a refresh token (scope: `user-top-read`).
-4. Paste `SPOTIFY_CLIENT_ID`, `SPOTIFY_CLIENT_SECRET`, `SPOTIFY_REFRESH_TOKEN` into `.env`.
-
-If Spotify is not configured, the DJ uses `settings.json` + `taste_seeds.json`.
-
 ## API

 | Method | Path | Description |
@@ -0,0 +1,96 @@
+# Taste from Spotify screenshots (no API keys)
+
+Live Ozan Radio does **not** connect to Spotify. Every track is generated fresh — we never play catalog music.
+
+Your taste lives in two files the DJ reads on every generate:
+
+| File | Purpose |
+|------|---------|
+| `settings.json` | Broad profile — genres, mood, instruments, tempo, references, avoid |
+| `taste_seeds.json` | Concrete anchors — artists, playlists, albums, track vibes |
+
+**Easiest setup:** open this repo in **Cursor** (or Claude Code), paste Spotify screenshots, and ask the agent to update those two files.
+
+---
+
+## What to screenshot
+
+Capture anything that shows your real listening — the more context, the better:
+
+1. **Spotify Home** — Daily Mixes, Jump back in, Made For you
+2. **Your Library** — playlists, artists, albums (scroll if needed)
+3. **Liked Songs** or a playlist you actually use
+4. **Now Playing** — optional, shows current mood
+
+Two or three screenshots usually enough. PNG from desktop app or phone both work.
+
+---
+
+## Cursor prompt (copy-paste)
+
+Attach your screenshots to the chat, then paste:
+
+```
+I'm setting up Live Ozan Radio (this repo). No Spotify API — taste only.
+
+Look at my Spotify screenshots and update:
+1. settings.json → taste section (summary, genres, mood, instruments, tempo_bpm, references, avoid)
+2. taste_seeds.json → genres, artists, playlists, and 5–15 track/album seeds with short "vibe" lines
+
+Rules:
+- Infer crossover taste the DJ can compose toward — not a copy of any one artist
+- settings.json = creative direction for Lyria; taste_seeds.json = concrete reference anchors
+- Never suggest playing Spotify URLs — generation only
+- Keep listener name as me unless I say otherwise
+- Show me a short summary of what you inferred before writing files
+```
+
+The agent edits the JSON files. Restart the server (or just generate — settings reload every request).
+
+---
+
+## Claude Code / pi / other agents
+
+Same flow: clone repo, add screenshots, use the prompt above. Entry point: `AGENTS.md` and `.cursor/skills/ozan-radio/SKILL.md`.
+
+---
+
+## Manual edit (no AI)
+
+Edit `settings.json` and `taste_seeds.json` by hand. See `taste_seeds.json` for the track object shape:
+
+```json
+{
+  "title": "Garip",
+  "artists": ["Altın Gün"],
+  "album": "Garip",
+  "vibe": "Anadolu psych-folk, bağlama, fuzzy warmth, hypnotic groove"
+}
+```
+
+---
+
+## Friends / group taste
+
+For a shared WhatsApp vibe (everyone's links and references), paste the **chat excerpt + screenshots** into Cursor:
+
+```
+Our friend group shares this taste (chat + screenshots attached).
+Update settings.json for a crossover station everyone would enjoy on repeat.
+Bias: [e.g. Anadolu psych, Soap&Skin ether, Mercan Dede, dubtronica]
+```
+
+Commit `settings.json` + `taste_seeds.json` if you want the profile in git. Keep `.env` private (API keys only).
+
+---
+
+## Example profile (checked in)
+
+The repo ships **Ozan's** profile as a reference — do not need to match it; replace with your own.
+
+| File | What it shows |
+|------|----------------|
+| [`settings.json`](../settings.json) | Full DJ direction — genres, mood, instruments, tempo, references, avoid |
+| [`taste_seeds.json`](../taste_seeds.json) | Concrete anchors — artists, playlists, albums, track vibes |
+
+Built from Spotify screenshots: Altın Gün, Baaba Maal, Thievery Corporation, islandman, Blanco White, ethnic chill, dubtronica, Anadolu psych.
@@ -390,7 +390,7 @@
    <h1>Ozan Radio</h1>
    <button type="button" class="icon-btn" id="settingsBtn" title="Settings" aria-label="Settings">⚙</button>
  </div>
-  <p class="tagline">No Spotify playback. No catalog tracks. DeepSeek picks the vibe — Google Lyria 3 composes it fresh.</p>
+  <p class="tagline">No catalog tracks. Taste from settings.json — set yours via Cursor + Spotify screenshots.</p>

  <section class="dashboard" id="dashboard">
    <div class="stat-card">
@@ -484,7 +484,7 @@
  </section>

  <div class="stack">
-    DJ: DeepSeek · Music: Google Lyria 3 · Taste: settings.json<br>
+    DJ: DeepSeek · Music: Lyria 3 · Taste: screenshots → Cursor → settings.json<br>
    Optional live layer: Magenta RealTime 2 on Apple Silicon
  </div>
 </div>
@@ -8,7 +8,7 @@ where = ["src"]
 [project]
 name = "live-ozan-radio"
 version = "0.1.0"
-description = "AI radio — DeepSeek DJ + Google Lyria 3, tuned to your Spotify taste"
+description = "AI radio — DeepSeek DJ + Google Lyria 3, taste from settings and screenshot seeds"
 requires-python = ">=3.11"
 dependencies = [
    "google-genai>=1.0",
@@ -1,18 +1,20 @@
 {
  "listener": "ozan",
+  "profile_note": "Example taste profile (Ozan / bozanbozkurt Spotify library, built from screenshots in Cursor). Copy this file structure for your own station — see docs/TASTE-FROM-SCREENSHOTS.md.",
  "station": "Live Ozan Radio",
  "taste": {
-    "summary": "Ethnic world dubtronica — global roots, dub space, and electronic groove.",
+    "summary": "Anadolu psychedelic rock meets ethnic dubtronica — Turkish psych, dark ether, Sufi electronic, Sahel warmth.",
    "genres": [
+      "anadolu psychedelic rock",
+      "turkish psychedelic rock",
      "ethnic world",
      "world dub",
      "dubtronica",
-      "dub",
-      "global electronica",
+      "dark ethereal indie",
+      "sufi electronic",
      "desert blues",
      "afro-dub",
      "middle eastern dub",
-      "balkan dub",
      "trip-hop dub"
    ],
    "mood": [
@@ -20,22 +22,26 @@
      "warm",
      "spacious",
      "late-night",
-      "sunset caravan",
-      "meditative but danceable"
+      "cinematic gothic",
+      "only lovers left alive",
+      "meditative but danceable",
+      "magnum opus slow burn"
    ],
    "instruments": [
+      "bağlama or saz",
+      "ney flute",
      "sub bass",
      "dub delay and spring reverb",
+      "fuzz guitar with middle eastern scales",
      "hand percussion",
-      "kora or oud",
-      "nose flute or melodica",
-      "tabla",
-      "djembe",
-      "muted guitar skank",
+      "darbuka",
+      "oud",
+      "melancholic piano",
+      "whispered vocal texture",
      "analog warmth"
    ],
-    "tempo_bpm": [82, 108],
-    "references": "Thievery Corporation meets Sahel dub; Baaba Maal warmth over a stepper bassline; Khruangbin haze with Lee Scratch Perry space; ethnic samples woven into dubtronica, not EDM.",
+    "tempo_bpm": [78, 102],
+    "references": "Altın Gün and Cem Karaca Anadolu psych; islandman and Kaya Project ethnic electronic; Thievery Corporation dub lounge; Baaba Maal Sahel; Blanco White melancholic ether; Shye Ben Tzur / Anoushka Shankar fusion; Jon Hopkins texture; Buddha Bar ethnic chill; Mungo's Hi Fi reggae-dub. Profile built from Spotify screenshots via Cursor — see docs/TASTE-FROM-SCREENSHOTS.md.",
    "avoid": [
      "big-room EDM drops",
      "four-on-the-floor house",
@@ -0,0 +1,10 @@
+{
+  "id": "80206d7a",
+  "title": "Whisper of the Dervish",
+  "mood": "hypnotic, late-night, slow-burn cinematic",
+  "dj_line": "This one's for the hours when the world falls away \u2014 a desert prayer for the dancing soul.",
+  "lyria_prompt": "Instrumental. Anadolu psychedelic rock meets Sufi electronic and dark cinematic ether. Tempo 85 BPM. Build slowly: start with a single, melancholic piano phrase and a distant, whispered vocal texture (wordless, breathy, processed with spring reverb). Layer in a melodic ney flute line in a minor maqam, doubled by a slightly distorted saz/ba\u011flama playing a hypnotic riff. Add a warm, deep sub bass pulse (dub style) and a slow, sparse darbuka pattern with hand percussion. Introduce a fuzz guitar with heavy spring reverb playing a slow, soaring solo in a Turkish psych scale (H\u00fcseyni or U\u015f\u015fak) around 1:00. Keep the mix spacious, with dub delay and analog warmth. The arrangement should feel like a slow-motion journey through a candlelit desert ruin. No drums until 0:45 \u2014 just a soft kick on the downbeat. End with the ney fading over a sustained bass note and piano chord.",
+  "lyrics": "[[A0]]\n[0.0:] (Haaaaaa...)\n[:] (Ooooooh...)\n[:] (Mmmmmmm...)\n[:] (Haaa-ooo-waaa...)\n[[B1]]\n[[A2]]\n[[C3]]\n[84.7:] (Haaaa...)\n[:] (Oooooh...)\n[[D4]]\n[[C5]]\n[118.6:] (Haaaaaaaah...)\n[[E6]]",
+  "file": "80206d7a_Whisper_of_the_Dervish.mp3",
+  "saved_at": "2026-06-07T13:47:36.171360+00:00"
+}
@@ -0,0 +1,10 @@
+{
+  "id": "e0701762",
+  "title": "Echoes of the Wandering Dervish",
+  "mood": "Hypnotic, warm, late-night cinematic slow burn",
+  "dj_line": "A slow-moving caravan across the Anatolian steppe, where the ney whispers secrets over a dubwise bassline.",
+  "lyria_prompt": "Turkish psychedelic rock meets Sufi electronic dub. Start with a deep, sub-heavy dub bassline at 85 BPM, pulsing with spring reverb and delay. Layer a melancholic piano phrase in a minor key, sparse and echoing. Introduce a ney flute melody, slow and hypnotic, floating over the bass. After 30 seconds, bring in a fuzz guitar playing a Middle Eastern scale phrase, heavily treated with reverb and analog warmth. Add darbuka and hand percussion in a slow, rolling pattern. Keep the arrangement spacious: no more than 4-5 elements at once. Build gradually, adding a whispered vocal texture (wordless, airy) in the second half. Avoid any sudden changes or drops. End with a slow fade on ney and bass. Analog warmth, tape saturation, and dub-style echo throughout. Instrumental with subtle vocal whispers.",
+  "lyrics": "[[A0]]\n[[A1]]\n[[A2]]\n[[B3]]\n[84.7:] (Haaaaaa...)\n[:] (Hoooooo...)\n[:] (Sssssssss...)\n[:] (Aaaaah...)\n[:] (Oooooh...)\n[[C4]]\n[118.6:] (Mmmmmm...)\n[:] (Haaaa...)\n[:] (Hoooo...)\n[[D5]]",
+  "file": "e0701762_Echoes_of_the_Wandering_Dervish.mp3",
+  "saved_at": "2026-06-07T13:44:08.752948+00:00"
+}
@@ -1,6 +1,6 @@
 {
  "index": 2,
-  "count": 4,
+  "count": 6,
  "tracks": [
    {
      "id": "b51b1cb2",
@@ -37,6 +37,24 @@
      "lyria_prompt": "Instrumental dubtronica track at 92 BPM in D minor. Start with a deep, sub-heavy bassline in a slow stepper rhythm. Layer a clean, plucked kora melody with long, spacey reverb and tape delay. Add a muted guitar skank playing offbeat chords with spring reverb. Hand percussion: djembe and shakers playing a relaxed, syncopated pattern. Introduce a melodica line with heavy echo, floating in the mix. Use analog warmth and subtle vinyl crackle. Build slowly, keeping the arrangement open and spacious. No vocals. Structure: intro with bass and percussion, add kora, then guitar, then melodica, then drop to just bass and percussion before returning with all elements. Fade out with delays.",
      "lyrics": "[[A0]]\n[[A1]]\n[[A2]]\n[[A3]]\n[[A4]]\n[[A5]]\n[[B6]]",
      "file": "eeebb429_Desert_Mirage.mp3"
+    },
+    {
+      "id": "e0701762",
+      "title": "Echoes of the Wandering Dervish",
+      "mood": "Hypnotic, warm, late-night cinematic slow burn",
+      "dj_line": "A slow-moving caravan across the Anatolian steppe, where the ney whispers secrets over a dubwise bassline.",
+      "lyria_prompt": "Turkish psychedelic rock meets Sufi electronic dub. Start with a deep, sub-heavy dub bassline at 85 BPM, pulsing with spring reverb and delay. Layer a melancholic piano phrase in a minor key, sparse and echoing. Introduce a ney flute melody, slow and hypnotic, floating over the bass. After 30 seconds, bring in a fuzz guitar playing a Middle Eastern scale phrase, heavily treated with reverb and analog warmth. Add darbuka and hand percussion in a slow, rolling pattern. Keep the arrangement spacious: no more than 4-5 elements at once. Build gradually, adding a whispered vocal texture (wordless, airy) in the second half. Avoid any sudden changes or drops. End with a slow fade on ney and bass. Analog warmth, tape saturation, and dub-style echo throughout. Instrumental with subtle vocal whispers.",
+      "lyrics": "[[A0]]\n[[A1]]\n[[A2]]\n[[B3]]\n[84.7:] (Haaaaaa...)\n[:] (Hoooooo...)\n[:] (Sssssssss...)\n[:] (Aaaaah...)\n[:] (Oooooh...)\n[[C4]]\n[118.6:] (Mmmmmm...)\n[:] (Haaaa...)\n[:] (Hoooo...)\n[[D5]]",
+      "file": "e0701762_Echoes_of_the_Wandering_Dervish.mp3"
+    },
+    {
+      "id": "80206d7a",
+      "title": "Whisper of the Dervish",
+      "mood": "hypnotic, late-night, slow-burn cinematic",
+      "dj_line": "This one's for the hours when the world falls away \u2014 a desert prayer for the dancing soul.",
+      "lyria_prompt": "Instrumental. Anadolu psychedelic rock meets Sufi electronic and dark cinematic ether. Tempo 85 BPM. Build slowly: start with a single, melancholic piano phrase and a distant, whispered vocal texture (wordless, breathy, processed with spring reverb). Layer in a melodic ney flute line in a minor maqam, doubled by a slightly distorted saz/ba\u011flama playing a hypnotic riff. Add a warm, deep sub bass pulse (dub style) and a slow, sparse darbuka pattern with hand percussion. Introduce a fuzz guitar with heavy spring reverb playing a slow, soaring solo in a Turkish psych scale (H\u00fcseyni or U\u015f\u015fak) around 1:00. Keep the mix spacious, with dub delay and analog warmth. The arrangement should feel like a slow-motion journey through a candlelit desert ruin. No drums until 0:45 \u2014 just a soft kick on the downbeat. End with the ney fading over a sustained bass note and piano chord.",
+      "lyrics": "[[A0]]\n[0.0:] (Haaaaaa...)\n[:] (Ooooooh...)\n[:] (Mmmmmmm...)\n[:] (Haaa-ooo-waaa...)\n[[B1]]\n[[A2]]\n[[C3]]\n[84.7:] (Haaaa...)\n[:] (Oooooh...)\n[[D4]]\n[[C5]]\n[118.6:] (Haaaaaaaah...)\n[[E6]]",
+      "file": "80206d7a_Whisper_of_the_Dervish.mp3"
    }
  ]
 }
@@ -10,24 +10,20 @@ from ozan_radio.dj import DeepSeekDJ
 from ozan_radio.lyria import LyriaEngine
 from ozan_radio.queue import RadioQueue
 from ozan_radio.server import app
-from ozan_radio.spotify import SpotifyTaste
 from ozan_radio.taste import load_taste_seeds


 async def generate_one() -> None:
    """CLI: generate a single track and print the result."""
    cfg = Config.from_env()
-    taste = await SpotifyTaste(cfg).fetch_taste()
-    seeds = None if taste else load_taste_seeds()
-    if taste:
-        print(f"Taste: {taste.summary}\n")
-    elif seeds:
+    seeds = load_taste_seeds()
+    if seeds:
        print(f"Taste seeds: {seeds.summary}\n")
    else:
-        print("No Spotify or seeds — DJ will freestyle.\n")
+        print("No taste_seeds.json — DJ uses settings.json only.\n")

    q = RadioQueue(cfg.output_dir)
-    plan = await DeepSeekDJ(cfg).plan_next(taste, q.recent_titles, seeds)
+    plan = await DeepSeekDJ(cfg).plan_next(q.recent_titles, seeds)
    print(f"DJ: {plan.dj_line}")
    print(f"Title: {plan.title}")
    print(f"Prompt: {plan.lyria_prompt}\n")
@@ -15,9 +15,6 @@ class Config:
    deepseek_base_url: str
    deepseek_api_key: str
    deepseek_model: str
-    spotify_client_id: str
-    spotify_client_secret: str
-    spotify_refresh_token: str
    radio_host: str
    radio_port: int
    output_dir: Path
@@ -37,9 +34,6 @@ class Config:
            ),
            deepseek_api_key=os.getenv("DEEPSEEK_API_KEY", ""),
            deepseek_model=os.getenv("DEEPSEEK_MODEL", "deepseek-chat"),
-            spotify_client_id=os.getenv("SPOTIFY_CLIENT_ID", ""),
-            spotify_client_secret=os.getenv("SPOTIFY_CLIENT_SECRET", ""),
-            spotify_refresh_token=os.getenv("SPOTIFY_REFRESH_TOKEN", ""),
            radio_host=os.getenv("RADIO_HOST", "127.0.0.1"),
            radio_port=int(os.getenv("RADIO_PORT", "8787")),
            output_dir=output,
@@ -54,10 +48,3 @@ class Config:
        if not self.deepseek_api_key:
            raise RuntimeError("DEEPSEEK_API_KEY is required for the DJ brain")

-    @property
-    def spotify_configured(self) -> bool:
-        return bool(
-            self.spotify_client_id
-            and self.spotify_client_secret
-            and self.spotify_refresh_token
-        )
@@ -7,7 +7,6 @@ from dataclasses import dataclass
 import httpx

 from ozan_radio.config import Config
-from ozan_radio.spotify import TasteProfile
 from ozan_radio.settings import ListenerSettings, load_settings
 from ozan_radio.taste import TasteSeeds

@@ -29,7 +28,7 @@ DJ_SYSTEM = """You are the DJ for Live Ozan Radio — a personal AI station that
 catalog music. Every track is generated fresh by Google Lyria 3 based on your prompts.

 Your job:
-1. Read the listener's Spotify taste (if provided).
+1. Read settings.json and taste_seeds.json (listener profile from screenshots or manual edits).
 2. Pick a mood, tempo, and genre blend that feels like a natural next track.
 3. Write a Lyria prompt that produces a 1–2 minute instrumental or vocal track.
 4. Keep variety — don't repeat the same vibe twice in a row.
@@ -121,7 +120,6 @@ class DeepSeekDJ:

    async def plan_next(
        self,
-        taste: TasteProfile | None,
        recent_titles: list[str],
        seeds: TasteSeeds | None = None,
        request: str | None = None,
@@ -132,12 +130,8 @@ class DeepSeekDJ:
            "Plan the next generated track for Live Ozan Radio.",
            f"Station taste:\n{self._taste_block()}",
        ]
-        if taste:
-            user_parts.append(f"Spotify taste: {taste.summary}")
-            if taste.top_genres:
-                user_parts.append(f"Genres: {', '.join(taste.top_genres[:8])}")
-        elif seeds:
-            user_parts.append(f"Taste seeds (no Spotify): {seeds.summary}")
+        if seeds:
+            user_parts.append(f"Taste seeds:\n{seeds.summary}")
        if recent_titles:
            user_parts.append(f"Already played (avoid repeating): {', '.join(recent_titles[-5:])}")
        if request:
@@ -13,7 +13,6 @@ from ozan_radio.config import Config
 from ozan_radio.dj import DeepSeekDJ
 from ozan_radio.lyria import LyriaEngine
 from ozan_radio.queue import RadioQueue
-from ozan_radio.spotify import SpotifyTaste
 from ozan_radio.chat_store import ChatStore
 from ozan_radio.dj import TrackPlan
 from ozan_radio.library import list_saved_songs
@@ -117,12 +116,11 @@ async def _compose_track(request: str | None = None, *, check_limit: bool = True
                    "message": f"Daily limit reached ({budget['max_per_day']} new songs)",
                    "budget": budget,
                }
-        taste = await SpotifyTaste(cfg).fetch_taste()
-        seeds = None if taste else load_taste_seeds()
+        seeds = load_taste_seeds()
        q = _get_queue()
        vibe = request or _chat.take_vibe_hint()
        plan = await DeepSeekDJ(cfg).plan_next(
-            taste, q.recent_titles, seeds, request=vibe or None
+            q.recent_titles, seeds, request=vibe or None
        )
        _set_generation(busy=True, phase="composing", title=plan.title)
        track = LyriaEngine(cfg).generate(plan)
@@ -137,7 +135,7 @@ async def _compose_track(request: str | None = None, *, check_limit: bool = True
            "status": "ok",
            "source": "generated",
            "track": np.__dict__ if np else None,
-            "taste_used": taste.summary if taste else (seeds.summary if seeds else None),
+            "taste_used": seeds.summary if seeds else None,
            "cost_usd": cost,
            "budget": budget,
        }
@@ -1,102 +0,0 @@
-from __future__ import annotations
-
-import base64
-from dataclasses import dataclass
-
-import httpx
-
-from ozan_radio.config import Config
-
-
-@dataclass
-class TasteProfile:
-    top_artists: list[str]
-    top_genres: list[str]
-    recent_tracks: list[str]
-    summary: str
-
-
-class SpotifyTaste:
-    """Read Ozan's listening taste — no playback, profile only."""
-
-    def __init__(self, config: Config) -> None:
-        self._config = config
-        self._token: str | None = None
-
-    async def _access_token(self) -> str:
-        if self._token:
-            return self._token
-        creds = f"{self._config.spotify_client_id}:{self._config.spotify_client_secret}"
-        auth = base64.b64encode(creds.encode()).decode()
-        async with httpx.AsyncClient(timeout=30) as client:
-            resp = await client.post(
-                "https://accounts.spotify.com/api/token",
-                headers={
-                    "Authorization": f"Basic {auth}",
-                    "Content-Type": "application/x-www-form-urlencoded",
-                },
-                data={
-                    "grant_type": "refresh_token",
-                    "refresh_token": self._config.spotify_refresh_token,
-                },
-            )
-            resp.raise_for_status()
-            self._token = resp.json()["access_token"]
-            return self._token
-
-    async def fetch_taste(self) -> TasteProfile | None:
-        if not self._config.spotify_configured:
-            return None
-
-        token = await self._access_token()
-        headers = {"Authorization": f"Bearer {token}"}
-
-        async with httpx.AsyncClient(timeout=30) as client:
-            artists_resp = await client.get(
-                "https://api.spotify.com/v1/me/top/artists",
-                headers=headers,
-                params={"limit": 10, "time_range": "medium_term"},
-            )
-            tracks_resp = await client.get(
-                "https://api.spotify.com/v1/me/top/tracks",
-                headers=headers,
-                params={"limit": 10, "time_range": "medium_term"},
-            )
-            artists_resp.raise_for_status()
-            tracks_resp.raise_for_status()
-
-        artists = artists_resp.json().get("items", [])
-        tracks = tracks_resp.json().get("items", [])
-
-        top_artists = [a["name"] for a in artists]
-        genres: list[str] = []
-        for artist in artists:
-            genres.extend(artist.get("genres", []))
-        # dedupe while preserving order
-        seen: set[str] = set()
-        top_genres = []
-        for g in genres:
-            if g not in seen:
-                seen.add(g)
-                top_genres.append(g)
-
-        recent_tracks = [
-            f"{t['name']} — {t['artists'][0]['name']}" for t in tracks if t.get("artists")
-        ]
-
-        genre_hint = ", ".join(top_genres[:6]) if top_genres else "eclectic"
-        artist_hint = ", ".join(top_artists[:5]) if top_artists else "varied"
-        summary = (
-            f"Listener leans toward {genre_hint}. "
-            f"Frequent artists: {artist_hint}. "
-            f"Recent favorites include {', '.join(recent_tracks[:3])}."
-            if recent_tracks
-            else f"Listener leans toward {genre_hint}. Frequent artists: {artist_hint}."
-        )
-
-        return TasteProfile(
-            top_artists=top_artists,
-            top_genres=top_genres[:12],
-            recent_tracks=recent_tracks,
-            summary=summary,
-        )
@@ -9,11 +9,14 @@ from pathlib import Path
 class TasteSeeds:
    genres: list[str]
    tracks: list[dict]
+    artists: list[str]
+    playlists: list[str]
+    albums: list[dict]
    summary: str


 def load_taste_seeds(repo_root: Path | None = None) -> TasteSeeds | None:
-    """Load taste_seeds.json from repo root — fallback when Spotify is offline."""
+    """Load taste_seeds.json — manual taste from screenshots or hand-edited seeds."""
    root = repo_root or Path(__file__).resolve().parents[2]
    path = root / "taste_seeds.json"
    if not path.exists():
@@ -26,19 +29,30 @@ def load_taste_seeds(repo_root: Path | None = None) -> TasteSeeds | None:

    genres = data.get("genres", [])
    tracks = data.get("tracks", [])
-    if not genres and not tracks:
+    artists = data.get("artists", [])
+    playlists = data.get("playlists", [])
+    albums = data.get("albums", [])
+    if not any([genres, tracks, artists, playlists, albums]):
        return None

-    track_hints = []
-    for t in tracks[:5]:
-        artists = ", ".join(t.get("artists", []))
-        vibe = t.get("vibe", "")
-        track_hints.append(f"{t.get('title', '?')} ({artists}) — {vibe}")
-
-    summary = "Seed taste profile. "
+    parts = ["Listener taste seeds (screenshot / manual profile)."]
+    if artists:
+        parts.append(f"Artists: {', '.join(artists[:12])}.")
+    if playlists:
+        parts.append(f"Playlists: {', '.join(playlists[:6])}.")
    if genres:
-        summary += f"Genres: {', '.join(genres[:8])}. "
-    if track_hints:
-        summary += "Reference vibes: " + "; ".join(track_hints)
+        parts.append(f"Genres: {', '.join(genres[:10])}.")
+    for album in albums[:4]:
+        parts.append(f"{album.get('title', '?')} — {album.get('vibe', '')}")
+    for t in tracks[:5]:
+        artist_str = ", ".join(t.get("artists", []))
+        parts.append(f"{t.get('title', '?')} ({artist_str}) — {t.get('vibe', '')}")

-    return TasteSeeds(genres=genres, tracks=tracks, summary=summary.strip())
+    return TasteSeeds(
+        genres=genres,
+        tracks=tracks,
+        artists=artists,
+        playlists=playlists,
+        albums=albums,
+        summary=" ".join(parts).strip(),
+    )
@@ -1,22 +1,72 @@
 {
  "listener": "ozan",
-  "notes": "Manual taste seeds when Spotify API is not wired. Add tracks from your library — DJ reads vibe, never replays catalog.",
+  "source": "Spotify screenshots → Cursor (no API). Re-run docs/TASTE-FROM-SCREENSHOTS.md when your library shifts.",
+  "playlists": [
+    "Buddha Bar Lounge - Ethnic Chill",
+    "Reggelectro and Dubtronica",
+    "Sci-Fi Scapes & Café Africa",
+    "Liked Songs",
+    "Your Top Songs 2025"
+  ],
+  "artists": [
+    "Altın Gün",
+    "Baaba Maal",
+    "Thievery Corporation",
+    "islandman",
+    "Cem Karaca",
+    "Kaya Project",
+    "Blanco White",
+    "Shye Ben Tzur",
+    "Anoushka Shankar",
+    "Jon Hopkins",
+    "Glass Beams",
+    "Francis Bebey",
+    "Mungo's Hi Fi",
+    "Matthew Halsall"
+  ],
+  "albums": [
+    { "title": "Garip", "artist": "Altın Gün", "vibe": "Anadolu psych-folk, bağlama, fuzzy warmth" },
+    { "title": "Estuaire", "artist": "Ablaye Cissoko", "vibe": "kora, West African elegance, spacious" },
+    { "title": "Mirage", "artist": "Glass Beams", "vibe": "psychedelic instrumental, Middle Eastern tint" },
+    { "title": "African Electronic Music 1975-1982", "artist": "Francis Bebey", "vibe": "early African electronic, playful hypnotic" }
+  ],
  "genres": [
-    "ethnic world",
+    "anadolu psychedelic rock",
+    "ethnic chill",
    "dubtronica",
    "world dub",
-    "west african",
-    "afro-folk",
-    "desert blues"
+    "trip-hop",
+    "desert blues",
+    "afro-dub",
+    "atmospheric electronic",
+    "global fusion jazz"
  ],
  "tracks": [
    {
      "title": "Boboyillo",
      "artists": ["Baaba Maal", "Rougi"],
      "album": "Being",
-      "added": "2023-07-16",
-      "duration_sec": 245,
-      "vibe": "Sahel warmth, rolling desert pulse, call-and-response vocals, acoustic strings and hand percussion, spiritual but danceable"
+      "vibe": "Sahel warmth, rolling desert pulse, call-and-response, spiritual but danceable"
+    },
+    {
+      "title": "Olalla",
+      "artists": ["Blanco White"],
+      "vibe": "melancholic indie folk, Spanish ether, intimate late-night"
+    },
+    {
+      "title": "Daily Mix 01",
+      "artists": ["BALTHVS", "Pentagram", "Cem Karaca"],
+      "vibe": "Turkish rock and psych crossover"
+    },
+    {
+      "title": "Daily Mix 03",
+      "artists": ["Thievery Corporation", "Télépopmusik"],
+      "vibe": "downtempo trip-hop, lounge space, dubbed edges"
+    },
+    {
+      "title": "Daily Mix 04",
+      "artists": ["Shye Ben Tzur", "Anoushka Shankar"],
+      "vibe": "Indian classical fusion, devotional texture, world spiritual"
    }
  ]
 }