| Command | What it tests |
|---|
npm test | Full suite — all tests |
npm run test:meaning | Communicative behavior guardrails |
npm run test:determinism | Hash-based regression tests |
npm run bench:rtf | Real-time factor benchmark |
npm run smoke | End-to-end smoke test |
Primary docs live in packages/voice-engine-dsp/docs/:
| Document | Contents |
|---|
STREAMING_ARCHITECTURE.md | Causal processing model, snapshot/restore, state management |
MEANING_CONTRACT.md | Prosody behavior specification — what the engine promises |
DEBUGGING.md | How to diagnose pitch issues, event ordering problems, and style drift |
Reference_Handbook.md (root) | Full API reference and concept guide |
PERF_CONTRACT.md (root) | Real-time factor and latency guarantees |
| Package | Path | Contents |
|---|
| voice-engine-dsp | packages/voice-engine-dsp/ | Core DSP, streaming prosody engine, meaning tests, determinism tests, RTF benchmarks |
| Aspect | Detail |
|---|
| Data touched | In-memory float arrays (audio samples), configuration objects |
| Data NOT touched | No file system writes, no network, no telemetry |
| Permissions | None — pure computation library with no I/O side effects |
| Network | None — fully offline |
See SECURITY.md for vulnerability reporting.