Dataview:
list from [[]] and !outgoing([[]])Audiobook download and stream app
- Artists I want:
- All my current downloaded albums
- Sections for:
- Jazz
- Jazz Vocalists
- Jazz study (covers of songs I like)
- Laufey
- Sarah Kang
- Cody Fry
- Beatles
- Jobim
- Support playlists, like:
- Genres
- Albums
- Artists
- I want the feature of showing how long playlists are
- Track playtime stats
🧲 Published
Deployment:
GitHub:
🎟 Features
Existing
Todo
🧾 Project Description
Blurt
Official
📂 Project Logs
🔗 -> Links
Resources
- Put useful links here
Connections
- Link all related words
LLM Context
### Project Goal
To create a dedicated, offline, ad-free music player (using an old Android phone) that provides **smart curation, setlists, and on-device recommendations** based on a pre-processed local library. This approach is intended to be a **non-trivial programming exercise** by creating local functionality that bypasses typical cloud services.
---
### Hardware Allocation
* **PC (Linux/Windows/macOS):** Used for bulk processing, high-resource tasks, and serving as the primary library source.
* **Old Android Phone (Target Device):** Used exclusively for offline playback, user interaction, and real-time (though lightweight) recommendation inference.
---
### Core Functional Components
#### 1. PC: Library & Training Server (Offline Pre-Processing)
* **Task:** **Acquisition, Curation, and Feature Extraction.**
* **Implementation:** Custom-written script (e.g., Python using **Mutagen** and **Librosa**).
* **Output:**
* **Clean Music Files:** Standardized metadata (Artist, Title, Album) written directly into the MP3/FLAC files.
* **Feature File (Embeddings):** A static, pre-computed data file (e.g., CSV or JSON) containing the numerical features (e.g., **BPM, Energy, Valence**) and **Song Embeddings** for the entire library. This is the **Content-Based Filtering (CBF)** training output.
#### 2. Android Phone: Player & Inference Client (Offline Interaction)
* **Task:** **Playback, Synchronization, Inference, and History Logging.**
* **Implementation:** A **forkable, modern Android Audio Player** (e.g., **Audiofy** or **OuterTune**) using **Kotlin** and **Media3**.
* **Data Input:** The pre-computed **Feature File** and the clean music files synced from the PC.
* **Key Features to Implement:**
* **On-Device CBF Recommendation Engine:** Uses the pre-computed features to perform fast, local vector calculations (e.g., **cosine similarity**) to generate recommendations (e.g., "Similar to this Song," "Similar to this Artist").
* **Custom Setlist Feature:** A unique data model and UI (separate from standard playlists) for structured listening.
* **Listening History/Metrics:** Local logging of play count, last played time, skips, etc., to feed back into the CBF and rule-based (e.g., "Discover Infrequently Listened") recommenders.
---
### Architectural Requirement
* The entire recommendation and playback process on the Android phone must be **fully offline** and **on-device**.
* The PC and Phone communicate only during **periodic synchronization** (for new music and updated feature files).