memp3

AI memory encoded in sound. Lossless, fast, and hardware-adaptive.

memp3 encodes text into FLAC audio files with Reed-Solomon error correction, stores them in SQLite, and provides semantic search via MCP for Claude Desktop.

Architecture

Text  -->  zlib compress  -->  Reed-Solomon ECC  -->  Frequency encoding (200-4000Hz)
      -->  FLAC compress  -->  SQLite blob storage

Search: FTS5 (word match) + FastEmbed ONNX + sqlite-vector (semantic similarity)

Everything lives in one SQLite database per user. No external services, no separate index files.

Performance

Operation	Time
Store memory	67ms
Retrieve memory	15ms
Word search (FTS5)	<1ms
Semantic search	150ms
MCP server cold start	127ms

Installation

pip install -e .

# With semantic search support
pip install -e ".[search]"

Claude Desktop Integration

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "memp3": {
      "command": "/path/to/venv/bin/python",
      "args": ["-u", "-m", "memp3.mcp"]
    }
  }
}

MCP Tools

Tool	Description
store_memory	Encode text into FLAC audio. Returns memory ID
retrieve_memory	Decode audio back to text by UUID
search_memories	FTS5 word-level search (any word matches)
semantic_search	Meaning-based search via embeddings
list_memories	List all stored memories
delete_memory	Delete a memory by UUID

CLI

# Store and retrieve
memp3 encode "Meeting at 3pm in Thane" --tags "meeting,thane"
memp3 decode <memory-id>

# Search
memp3 search "meeting"
memp3 semantic-search "schedule appointment"

# Manage
memp3 list
memp3 delete <memory-id>
memp3 info <memory-id>
memp3 stats

# Export
memp3 export-flac <memory-id>
memp3 export-wav <memory-id>

# Servers
memp3 mcp          # MCP stdio server (Claude Desktop)
memp3 serve        # REST API server

REST API

# Start server
memp3 serve

# Create API key
memp3 create-key alice

# Use API
curl -X POST http://localhost:8000/memories \
  -H "Authorization: Bearer memp3_xxxx" \
  -H "Content-Type: application/json" \
  -d '{"content": "Meeting at 3pm", "tags": "meeting"}'

curl http://localhost:8000/memories?query=meeting \
  -H "Authorization: Bearer memp3_xxxx"

How It Works

Encoding Pipeline

"Hello World"
    |
    v
UTF-8 bytes --> zlib compress --> Reed-Solomon ECC (14% redundancy)
    |
    v
CRC32 header (magic + version + length + checksum)
    |
    v
Frequency mapping: each byte -> 200-4000Hz tone (256 slots)
    |
    v
48kHz audio signal with Hann windowing (reduces spectral leakage)
    |
    v
FLAC lossless compression --> SQLite blob

Search

• FTS5: Word-level matching. "Lakshya wedding" finds "Lakshya is getting married" because "Lakshya" matches.
• Semantic: FastEmbed (ONNX) generates 384-dim embeddings, sqlite-vector does cosine similarity. "motorcycle" finds "bike" (0.837 score).

Error Correction

Reed-Solomon RS(255,223) recovers the original text even if 5% of audio samples are corrupted.

Tech Stack

Component	Technology
Audio encoding	numpy, scipy, soundfile
Error correction	reedsolo (Reed-Solomon)
Storage	SQLite (WAL mode, FLAC blobs)
Word search	SQLite FTS5
Semantic search	FastEmbed (ONNX) + sqlite-vector
MCP server	Lightweight JSON-RPC over stdio
REST API	FastAPI
CLI	Typer

Multi-tenant SaaS

Per-user isolation with API key auth:

/data/memp3/users/
  alice/index.db    <-- all of Alice's memories
  bob/index.db      <-- all of Bob's memories

Delete user = delete their folder. No shared database.

License

MIT