VoxReason: Auditing Source-Grounded Speech Plans Before Synthesis

Mengzhe Geng

cs.SDSep 2, 2026

VoxReason: Auditing Source-Grounded Speech Plans Before Synthesis

Authors: Mengzhe Geng

Abstract

Speech systems increasingly infer how an utterance should be delivered from context, but a plausible delivery plan may not be supported by the input. VoxReason is a small public benchmark and verifier for testing this failure before waveform synthesis. Each of its 100 cases fixes the utterance, provides derived records that name the source emotion and intensity, and changes one licensed cue. A system must cite the record for its delivery decision and update only the plan fields associated with the edit. We use deterministic verifier references, not a model leaderboard. This holdout excludes every emotion and intensity combination observed in training. A prior-only predictor achieves 0.958 accuracy across plan fields but never changes its plan consistently after a cue edit. This contrast shows that plan agreement does not demonstrate source grounding. The released suite provides an auditable measurement layer for screening structured speech plans before synthesis. It is limited to derived records, not audio inputs.

Explore similar work

Beyond Prompt Adherence: Auditing Attribute-Level Voice Control in Speech Generation
Aug 1, 2026Xianhao Zhou, Jianghao WuAdherence Preservation
AVERT: Audio-Verified Adjudication for Spoken Dialogue State Tracking
Sep 1, 2026Chunggi Lee, Hanspeter PfisterDialogue Adjudication
SpeakerLLM: A Speaker-Specialized Audio-LLM for Speaker Understanding and Verification Reasoning
May 14, 2026KiHyun Nam, Jungwoo Heo, Siu Bae +2Speaker Verification Performance Speech Evaluation

Aug 1, 2026 · cs.SDJ/K move · Enter open · S save

Beyond Prompt Adherence: Auditing Attribute-Level Voice Control in Speech Generation

Xianhao Zhou, Jianghao Wu

School of Mechanical and Electrical Engineering, UESTC, Chengdu, China · Faculty of Information Technology, Monash University, Melbourne, Australia

Natural-language descriptions have become a flexible interface for controlling generated speech. Existing evaluations largely assess whether an output matches a prompt, but prompt matching alone does not reveal whether characteristics outside the intended change remain stable. We examine this distinction through a controlled paired audit of three speech-generation systems: CosyVoice3, VoxCPM2, and Fish-Speech-S2. The evaluation contains 5,940 outputs spanning six reference speakers, ten texts, three random seeds, and eleven conditions. Using acoustic, prosodic, content, and speaker measurements, we find that responses in the expected target direction are frequently accompanied by changes outside descriptor-specific signal-level target sets. This pattern remains among outputs whose target response exceeds baseline seed variation, and the accompanying changes differ substantially across systems. We further introduce VoDER-Cal, a training-free candidate selector that retains sufficiently strong target responses while favoring smaller off-target deviations. A three-candidate pool raises the joint success rate from 4.8% under single-sample direct generation to approximately 14% for all candidate-selection policies. Within the matched three-candidate budget, VoDER-Cal reduces held-out off-target deviation from 0.344 under target-only selection to 0.276 and improves listener-rated preservation. Preservation-sensitive evaluation therefore complements prompt-adherence evaluation, while candidate reranking offers a practical inference-time improvement. Code, configuration files, and analysis scripts are available at https://github.com/intelland/VoDER

Adherence Preservation

Open paper Ask about it PDF

Abstract

Explore similar work

Beyond Prompt Adherence: Auditing Attribute-Level Voice Control in Speech Generation

AVERT: Audio-Verified Adjudication for Spoken Dialogue State Tracking

SpeakerLLM: A Speaker-Specialized Audio-LLM for Speaker Understanding and Verification Reasoning