NVIDIA's Voice Memory: Optimizing Speech Recognition via Editable Memory Files
nvidia · hf · 2026-07-31
NVIDIA introduced Voice Memory, an inference-only scheme for agentic speech recognition utilizing a listener-thinker architecture:
- Listener: Reads a per-domain memory.md file at stream time to decide whether to correct the current recognition hypothesis.
- Thinker: Asynchronously updates the memory file, accepting edits only if they strictly improve a held-out score.
This design requires no weight changes, keeping learned skills auditable and portable. Experiments show that unconstrained generative error correction often over-corrects, breaking correct tokens (up to 64% of the time on financial news), whereas Voice Memory reduces this rate to 35%. Across ten domains, the scheme lowers the weighted word error rate (WWER) from 8.36% to 7.52% without adding inference parameters.
More from Research
- One ancient fold, many jobs: how the immunoglobulin domain evolved into antibodies, MHC and titin — arjunrajlab · 2026-10-16
- Universal forgery attack breaks Chinese PQC signature Origami using only the public key — matthew_d_green · 2026-10-16
- How do you attribute agent gains: model training or infrastructure? — jrxna · 2026-10-16
- First-ever paper on multi-agent swarm safety lands NeurIPS spotlight — LauraRuis · 2026-10-16
- EDEN: anchored embedding noise boosts RLVR exploration beyond policy rollouts — Daijiao Liu · 2026-10-16
- Autonomous vehicle demo shows car reversing to break narrow-road deadlock, with a robotic arm to shift gears — sanjeevs_iitr · 2026-10-16