WaveCommitments: Private and Verifiable Speech-Agent Marketplaces

Oguzhan Baser; Kaan Kale; Sriram Vishwanath; Sandeep Chinchali

cs.CLJul 16, 2026

WaveCommitments: Private and Verifiable Speech-Agent Marketplaces

Authors: Oguzhan Baser, Kaan Kale, Sriram Vishwanath, Sandeep Chinchali

Organizations: †The University of Texas at Austin, Austin, TX, USA · ‡Georgia Institute of Technology, Atlanta, GA, USA

Abstract

Modern speech systems are evolving into ecosystems of autonomous speech agents that interact, negotiate, and trade assets such as embeddings, spectrograms, or raw data. In such markets, sellers cannot reveal assets for evaluation due to privacy constraints, and buyers cannot price what they cannot test. Thus, we introduce WaveCommitments, an end-to-end design that couples multivariate polynomial commitments with confidential asset valuation via Data Shapley inside a trusted execution environment. It enables selective openings and tamper-evident provenance to prove that a delivered asset matches the valuation while hiding everything else. Our extensive experiments across three standard datasets and various tasks with state-of-the-art encoders, including Whisper, reveal that we speed up the strongest baselines by up to 18.1x while reducing their proof sizes by up to 43.2x. We also analyze how a speaker's value shifts across tasks and encoders, revealing agent-level market prices.

Figures & tables

Figure 1: WaveCommitment Architecture: Imagine a SAM where autonomous agents trade proprietary speech knowledge (red) , such as embeddings. Seller $s$ cannot reveal raw utterance $u$ , and buyer $a$ cannot trust unseen embeddings $E^{s}$ or determine their value $V$ for its model $H^{a}$ . Hence, they both upload their assets to a secure enclave which computes the valuations for each pair confidentially, seals those and the referenced assets into a single digest $C$ to publish in a public bulletin (green) . After $a$ and $s$ handshake based on $V$ , $a$ buys $E^{s}$ and uses $C$ to validate it without exposing other assets.

Figure 2: How to selectively verify speech tensors? WaveTrees (c) organize assets in native time-frequency geometry and return a short proof for the region of interest, unlike Merkle (a) and Verkle (b) trees with larger overhead (outlined boxes). Ours can verify whether a speaker says a phrase in a time window while copyrighted music on background frequencies remains hidden.

Figure 3: Per-Speaker DS for SAM Pricing: For each speaker, we show DS distributions over utterance-level embeddings across three buyer heads (color) and three feature extractors (hatch), computed confidentially in the TEE. This reveals who to buy from and which embedding space best serves each task.

	Metric	MT	UCT	WT (Ours)
TIMIT	Commit Size ( $\downarrow$ )	32B	32B	32B
	Proof Size ( $\downarrow$ )	4280B	480B	128B
	Commit Time ( $\downarrow$ )	12.2 $\pm$ 2.6ms	409 $\pm$ 8.7ms	32.6 $\pm$ 2.8ms
	Verify Time ( $\downarrow$ )	$\approx$ 1ms	25.2 $\pm$ 2.2ms	5.6 $\pm$ 1.1ms
IEMOCAP	Commit Size ( $\downarrow$ )	32B	32B	32B
	Proof Size ( $\downarrow$ )	6480B	672B	157B

Table 1: Do SAMs validate the assets fast enough to trade at scale?

Privacy-Preserving Speech Processing

See all 24 papers

TensorCommitments: A Lightweight Verifiable Inference for Large Language Models
Feb 13, 2026Oguzhan Baser, Elahe Sadeghi, Eric Wang +5LLM Inference LLM Security
Electrical and Computer Engineering, The University of Texas at Austin · Theseus AI Labs · Electrical and Computer Engineering, McGill University +1
The Watermark Shortcut: How Provenance Marking Sabotages Audio Deepfake Detection
Jun 22, 2026Nicolas M. Müller, Pascal DebusAudio Watermarking Audio Deepfake Detection
Fraunhofer AISEC, Germany · Resemble AI
LISE : Listenable Interpretable Speaker Embeddings
Jun 19, 2026Xiaoliang Wu, Chongxin Gan, Ke Liu +2Audio Representation Learning Speaker Verification
University of Southampton, United Kingdom · The Hong Kong Polytechnic University, Hong Kong SAR, China · University of Edinburgh, United Kingdom

Feb 13, 2026J/K move · Enter open · S save

TensorCommitments: A Lightweight Verifiable Inference for Large Language Models

Oguzhan Baser, Elahe Sadeghi, Eric Wang, Nico Vergauwen+4

Electrical and Computer Engineering, The University of Texas at Austin · Theseus AI Labs · Electrical and Computer Engineering, McGill University+1

LLM Inference LLM Security Verifiable Outsourced Computation

Open paper AskPDF arXiv

Most large language models (LLMs) run on external clouds: users send a prompt, pay for inference, and must trust that the remote GPU executes the LLM without any adversarial tampering. We critically ask how to achieve verifiable LLM inference, where a prover (the service) must convince a verifier (the client) that an inference was run correctly without rerunning the LLM. Existing cryptographic works are too slow at the LLM scale, while non-cryptographic ones require a strong verifier GPU. We propose TensorCommitments (TCs), a tensor-native proof-of-inference scheme. TC binds the LLM inference to a commitment, an irreversible tag that breaks under tampering, organized in our multivariate Terkle Trees. For LLaMA2, TC adds only 0.97% prover and 0.12% verifier time over inference while improving robustness to tailored attacks by up to 48% over the best prior work requiring a verifier GPU.

Abstract

Privacy-Preserving Speech Processing

Explore similar work

TensorCommitments: A Lightweight Verifiable Inference for Large Language Models

The Watermark Shortcut: How Provenance Marking Sabotages Audio Deepfake Detection

LISE : Listenable Interpretable Speaker Embeddings