FAQ

General

Is Sound Suite really free?

Yes. Sound Suite is free for personal, educational, and non-commercial use under the Polyform Noncommercial License. You can use it to manage your own legal cases without any cost, subscription, or usage limits.

For commercial use by law firms or legal organizations, a commercial license is available. See the Pricing page for details.

What types of documents does it support?

Sound Suite processes PDF files. This covers the vast majority of legal filings, court records, briefs, motions, orders, and transcripts. PDFs can be:

  • Digital PDFs — Text is extracted directly (fast)
  • Scanned PDFs — Text is extracted via OCR using tesseract.js (slower but automatic)
  • Mixed PDFs — Pages are processed individually, using direct extraction or OCR as appropriate

Does my data leave my computer?

Not by default. When using local embedding providers (Transformers.js or Ollama), all processing happens on your machine. The MCP server only listens on localhost by default.

If you choose a cloud embedding provider (OpenAI or Anthropic), text chunks are sent to the provider's API for embedding generation. The AI Search feature also sends queries to your chosen AI provider.

How many documents can it handle?

Sound Suite has been tested with cases containing thousands of pages across dozens of documents. The SQLite database and LanceDB vector store are designed for local-scale workloads. Typical performance:

  • Small case (5-10 documents, ~100 pages): Indexes in minutes
  • Medium case (20-50 documents, ~1,000 pages): Indexes in 15-30 minutes
  • Large case (100+ documents, ~5,000+ pages): Indexes in 1-2 hours

Performance varies based on your hardware, embedding provider, and whether OCR is needed.

Setup & Requirements

What are the system requirements?

  • Node.js 18+ (required)
  • 4 GB RAM minimum, 8 GB recommended
  • macOS, Linux, or Windows
  • Ollama (optional, for local GPU-accelerated embeddings)
  • Redis (optional, for caching)

See the full System Requirements section.

Do I need a GPU?

No. Sound Suite works on CPU-only machines. However, a GPU significantly speeds up:

  • Local embedding generation (via Ollama)
  • OCR processing for scanned documents

Apple Silicon Macs perform well even without a discrete GPU due to the Neural Engine.

Can I use it offline?

Yes, when configured with local providers:

  • Embedding: Use transformers or ollama provider
  • Search: Direct Search and Analysis Tools work offline
  • AI Search: Requires an AI provider (cloud or local via Ollama)

Self-Hosting for Firms

Can multiple people access the same instance?

Yes. Sound Suite runs as a web application on your local network. Any device on the same network can access the dashboard at http://{your-ip}:3000. The Admin > General page shows your machine's network addresses.

For shared access, consider:

  • Enabling API Key authentication on the MCP server
  • Running behind a reverse proxy (nginx) for HTTPS
  • Using Docker for consistent deployment

Is it suitable for a law firm?

Sound Suite is designed for individual practitioners and small firms who want to self-host their document intelligence. Advantages for firms:

  • No cloud dependency — Meets strict data residency requirements
  • No per-user pricing — One instance serves everyone on the network
  • Full control — You own the server, database, and all data
  • MCP integration — Works with Claude Desktop and other AI tools your team already uses

For larger deployments, a commercial license includes support and priority updates.

Can I run it on a server?

Yes. Sound Suite works well on a dedicated server or NAS device:

# Production mode
npm run svc:start:prod

# Or with Docker
docker-compose up -d

The Docker deployment is recommended for server environments.

Data & Privacy

Where is my data stored?

All data stays on your machine:

Data Type Location
Database prisma/data/sound-suite.db (SQLite)
Vectors data/lancedb/ (LanceDB)
Exhibits public/exhibits/ (extracted images)
Logs logs/
Config .env file at project root

Can I export my data?

Yes. Your data is portable:

  • Database: Standard SQLite file — open with any SQLite tool
  • Vectors: LanceDB files — compatible with the LanceDB ecosystem
  • Documents: Your original PDFs are never modified; they remain in their original directories
  • Drafts: Export to .docx or .pdf from the Draft Editor

Can I back up my data?

Yes. Use the built-in backup commands:

# Create backup
npm run db:backup

# Restore from backup
npm run db:restore

See the Administration guide for details.

What happens if I uninstall?

Your original PDF files are never modified or moved. Uninstalling Sound Suite removes only the application files. The SQLite database and LanceDB vectors remain in the data directories until you delete them manually.

Technical

What embedding models are supported?

Sound Suite supports multiple embedding providers and models:

Provider Example Models Dimensions
Transformers.js all-MiniLM-L6-v2 384
Ollama qwen3-embedding:0.6b, nomic-embed-text 1024, 768
OpenAI text-embedding-3-small 1536
Anthropic claude-embed-1 1024

You can change providers at any time through Admin > Embedding, but existing documents will need re-indexing.

What is MCP?

The Model Context Protocol is an open standard for connecting AI assistants to external data sources and tools. Sound Suite's MCP server lets AI assistants like Claude Desktop, Cursor, and Windsurf directly query your case documents and run analysis tools.

Learn more in the MCP Tools Reference.

Can I add custom analysis tools?

The 14 built-in tools cover the most common legal analysis tasks. Custom tool development is on the roadmap. In the meantime, you can use the query_case_knowledge tool with detailed prompts to achieve custom analysis through AI.

How does OCR work?

Sound Suite uses tesseract.js (a JavaScript port of Tesseract OCR) to extract text from scanned pages. When a page has fewer characters than the density threshold (default: 50), the page image is rendered and sent through OCR. This is automatic and requires no configuration.