Voice identification (speaker recognition for multi-user)

[martinjms]

Summary

Speaker identification via voice embeddings / voiceprint enrollment for the voice channel. Enables per-user memories and auth in voice conversations.

References

• ARCHITECTURE.md Module 4 — User-aware memory — Voice identification