-
No external baseline (the strongest verified weakness). All comparators — "the original embeddings, Gaussian perturbation, a content-only representation, and Iterative Nullspace Projection (INLP)" (§5, p. 6), plus the ablations — are your own runs; there is no VoicePrivacy-style x-vector replacement or Chen et al. 2025 pseudo-speaker-distribution comparator, even though §2 describes Chen et al. as "closely related to the replacement component of MoIRAGE". We acknowledge the mitigations already present: validation-only INLP selection, a four-point σ sweep, best-operating-point reporting for baselines (Gaussian σ=1 matches the no-ECAPA linkage only at much worse utility, stated plainly), and strongest-observed-attacker reporting with the train/eval gap disclosed. Still, adding one external replacement baseline adapted to embeddings would materially strengthen the empirical claims.
-
No sensitivity analysis for MoIRAGE's constants. No robustness results are reported for τ=0.5, Huber δ=0.5, the privacy ramp (epochs 25→85), the 0.12× target-eigenvalue floor, or the 0.3 effective-rank threshold; and the attacker capacity sweep — "width {64, 256}, depth {2, 3}, and training budgets {400, 800, 1600} steps" — covers only the main learned families at K=20 (App H, p. 18). A small robustness table, or an explicit scope note, would suffice.
-
The promised effective-set-size reporting is absent. App G (p. 17) commits to "report the operating point as all available and provide the mean effective set size" when availability saturates, yet no such numbers appear anywhere. With a median of 50 chunks per participant, a full K=50 positive trial needs ≥100 recordings (supportable by well under half the cohort), so it is ambiguous whether, e.g., 0.854 (Table 4, K=50) is a fixed-K or an all-available number. The repeated plateau values (cosine 0.620 at K∈{20,30,50}; Bridge2AI cosine 0.575 and original 0.935 at K∈{5,20,50}) are consistent with saturation, but the paper never says so. The K=1→20 headline trend is unaffected. Please report the mean effective set size per reported K, plus one clarifying sentence.
-
The same-recording diagnostic is described but never reported. "A separate same-recording diagnostic receives two independent releases of the same recording" (App G, p. 17) and is said to be "reported separately" in App H, but no corresponding numbers appear in any table, figure, or prose (the cross-space results do appear, Tables 5 and 7). Please report the numbers or remove the description.
-
The Bridge2AI clinical-retention claim needs a pointer to App O. §6.4 (p. 8) states "Clinical information is also retained on Bridge2AI" without acknowledging that 10 of 12 auxiliary conditions decrease after transformation (App O / Table 9: muscle tension dysphonia 0.865→0.575, airway stenosis 0.904→0.713, PTSD 0.774→0.606). The sentence is true for the primary label (0.948 vs. 0.913, overlapping CIs), but one sentence pointing to App O would make the claim complete.
-
No confidence interval for the abstract's "close to chance". The abstract reports that the strongest identity-linkage AUC⋆ "decreases from 0.778 to 0.528, close to chance" (Abstract, p. 1), but the K=1 value has no interval anywhere, while K=20 is reported with one (0.811 [0.779,0.838], App I). Add the K=1 interval or soften the wording to "near chance".
-
Non-monotonic attacker columns without intervals or comment. The MoIRAGE attention-pooling column in Table 5 (p. 20) is non-monotonic (0.503→0.559→0.609→0.730→0.666→0.848→0.675), and DeepSets dips at K=50 (0.674→0.658), with no per-cell CIs and no comment on the 0.848→0.675 swing. Per-K retrained attackers plus saturation plus finite trials can produce this pattern (it is likely noise), but please add per-cell intervals or one commenting sentence.
-
The 0.840 vs. 0.811 gap is never reconciled. The 45-run fold-seed study reports "mean attacker AUC⋆ was 0.840 with a between-run standard deviation of 0.039" (§6.3, p. 8 / App I), i.e., 0.840±0.039, while the headline K=20 number is 0.811 (pooled TEST, Table 1). These are different aggregations (mean of per-run AUCs vs. AUC of pooled predictions), but the text never reconciles the 0.029 gap. One sentence would close it.
-
Unexplained "—" cells for the cross-space attacker at K=50. These appear in both Table 5 (p. 20, Original and MoIRAGE rows) and Table 7 (p. 23) with no stated reason; every other missing-cell pattern is protocol-explained except this one. Add one clause (presumably unsupported set size or protocol inapplicability).
-
Rounding inconsistency in the paired AUROC difference. "The paired AUROC difference is +0.075" (§6.2, p. 7), while the difference of the pooled Table 1 values is 0.643−0.567=0.076 — almost certainly rounding of unrounded underlying values. Note the rounding, or report the difference of the rounded quantities.
-
K=30 appears without motivation. It shows up only as a Table 5 column; §5 (p. 6) states "We mainly evaluate K∈{1,2,5,10,20}", and K=50 is motivated in §6.1/§7. Add one clause or drop the column.
-
Table 3 column swap. In Table 3 (p. 19), the −ECAPA and MoIRAGE columns conflict with three independent locations — §6.2 ("Elastic Net reaches an AUROC of 0.641 with MoIRAGE", p. 8; SVM MoIRAGE =0.652), Table 4 (MoIRAGE utility 0.643, −ECAPA 0.634), and App J ("For MoIRAGE, logistic-regression AUROC is 0.643", p. 19) — while Table 3 gives logistic regression −ECAPA 0.643 / MoIRAGE 0.634 (Elastic Net 0.641/0.627; SVM 0.652/0.644). This is very likely a paper-side column swap; it is minor because the correct numbers appear in three other locations. Swap the two column headers (or cells).
-
The Bridge2AI feature-extraction configuration is missing. The vectors are described only as "task-level acoustic feature vectors rather than WavLM embeddings" (App F, p. 17) — no feature set, dimensionality provenance, normalization, PCA status, or extraction software is named, unlike the fully specified UNMC pipeline (App A/E). As described, the replication cohort cannot be re-derived. One paragraph mirroring App A/E would fix this.
-
The checkpoint-selection feasibility thresholds have no values. "A checkpoint must satisfy clinical-utility, covariance-trace, effective-rank, and attack thresholds before it is considered feasible" (App D, p. 15), but the thresholds are never given values, so the selection rule — lowest worst-case validation AUC⋆ among feasible checkpoints — is not executable as stated. Please report the four values.
-
INLP's selection criterion is unstated. "32 directions satisfy the predefined utility-feasibility criterion" (App K, p. 20), but the criterion's value is absent, so the baseline's tuning fairness cannot be audited. State the criterion.
-
The unit "row" is never defined. It is used in both chunk and participant-day senses across App A/C/D/F/G (e.g., "One epoch is one full-batch generator update over the training rows of a fold", App D, p. 15); App D's "For fold 0 this contains 2,360 rows" matches participant-day arithmetic (86×28.2≈2,425) rather than chunk arithmetic (≈6,400 expected if rows = chunks at the mean of 75.1). Define the unit and reconcile the number.
-
Governance enhancements absent. No IRB protocol number is given ("an IRB-approved longitudinal study", Ethics Statement, p. 9), and there is no statement on whether derived embeddings or the frozen manifest are shareable, or under what terms/license. One clause each would close both gaps.
-
ECAPA is uncited at first use. §4 (p. 5) states "the identity-related representation is supervised using a frozen ECAPA speaker embedding" with no citation; Desplanques et al. 2020 appears only in App A. Cite it at first use in §4.
-
Novelty framing — keep it exactly as stated (§2, p. 3). The novelty claim is explicitly a combination claim — synthetic identity replacement, multi-observation adversarial training, and repeated-observation evaluation, brought together in a clinical embedding setting — and it is defensible as written: as §2 puts it, "these directions are typically studied separately." The closest prior art (Chen et al., 2025) is accurately characterized and cleanly differentiated (learned clinical embedding vs. resynthesized waveform), and no benchmark-SOTA claim is made anywhere in the paper. Please keep the claim framed exactly this way; the scoped, combination-level framing is the right one and reviewers should not push you toward a broader one.
-
Complete the VoicePrivacy citation chain (§2). The discussion currently cites the VoicePrivacy 2020 results overview and the 2024 challenge plan, but skips the 2022 results overview: Panariello et al., "The VoicePrivacy 2022 Challenge: Progress and Perspectives in Voice Anonymisation" (2024). Adding it in §2 completes the challenge-series chain (2020, 2022, 2024) and costs one line.
-
Anchor the linkage metrics to the linkability literature (§3 or §5). The linkage-metric apparatus — AUC⋆, EER, and TPR@FPR (e.g., "We additionally report EER," §5, p. 6) — is introduced without a linkability-metric citation. Nautsch et al., "Preserving privacy in speaker and speech characterisation" (2019), is the foundational speech-privacy/linkability reference for these quantities; a single citation in §3 or §5 will satisfy reviewers coming from the voice-privacy community.
-
Acknowledge the ADReSS/ADReSSo/ADReSS-M benchmark series (§5 or Appendix P). The standard dementia-recognition-from-speech benchmark series — ADReSS; ADReSSo (Luz et al., "Detecting cognitive decline using speech only: The ADReSSo Challenge," 2021); and ADReSS-M ("An Overview of the ADReSS-M Signal Processing Grand Challenge on Multilingual Alzheimer's Dementia Recognition Through Spontaneous Speech," 2024) — is neither cited nor acknowledged, even though the cited Woszczyk et al. (2024) evaluates on ADReSS/ADReSSo. The omission is defensible on threat-model grounds: at ∼1 recording per speaker, these corpora cannot support the multi-recording linkage protocol, whereas UNMC's median of 50 recordings per participant is genuinely better suited. Still, one acknowledging sentence in §5 or Appendix P making exactly this argument would preempt the obvious reviewer question. If space allows, the newer PROCESS Challenge may also be mentioned as an optional emerging benchmark.
-
Verify and tidy the Bridge2AI-Voice bibliography entry (References, p. 10). The entry prints DOI 10.13026/k81f-qr68, which resolves externally to the sibling DOI 10.13026/37yb-1t42 with identical title and authors — PhysioNet mints per-version DOIs, so a version alias is the likely explanation. Please verify that the printed string resolves as printed. The same entry also uses "et al." inside the bibliography (please give the full author list) and mis-capitalizes the dataset name as "Bridge2ai-voice" rather than "Bridge2AI-Voice" ("Bridge2AI-Voice: An ethically-sourced, diverse voice dataset linked to health information").
-
Dwork & Roth (2014) is acceptable, with an optional refinement (§2, p. 3). Citing Dwork & Roth (2014) for the composition phenomenon — "privacy can weaken when information from repeated releases is combined" — is a broad textbook citation, but it is tolerable as written because the §2 sentence is a general phenomenon statement rather than a specific technical claim. If space allows, a more specific composition reference could be considered; no change is required.
-
External-validity boundary is inherent, not a flaw — consider stating it explicitly (§5, p. 5; Appendix Q, p. 25). Because UNMC is private ("UNMC recordings cannot be redistributed by the authors," Appendix Q, p. 25), no same-benchmark comparator exists and the absolute numbers cannot be leaderboard-validated. Likewise, we found no published linkage or anonymization results on Bridge2AI-Voice, so the teacher-free replication appears to be the first privacy result on that cohort. As the paper itself notes, this is "an external development replication rather than a second confirmatory test" (§5, p. 5). All privacy/utility evidence therefore rests on a single group's pipeline; this is inherent to the datasets rather than a methodological flaw, but it bounds external validity. A single sentence in §2 or §5 making this boundary explicit would read as confidence rather than caveat.
-
Two further relevant works worth citing (§1/§2). Two externally identified works connect directly to your premise: Tayebi Arasteh et al., "The effect of speech pathology on automatic speaker verification: a large-scale study" (Scientific Reports, 2023), which provides large-scale evidence that pathological speech carries heightened, disorder-dependent re-identification risk — direct support for the §1 premise that clinical speech embeddings are identity-sensitive — and Tayebi Arasteh et al., "Federated Learning for Secure Development of AI Models for Parkinson's Disease Detection Using Speech from Different Languages" (Interspeech, 2023), on federated privacy-preserving clinical speech modeling. Consider citing both in §1/§2.
-
Give the effective-rank concept a reference (Appendix C, p. 14; Appendix N, p. 22). The effective-rank concept is used without a citation — "the entropy-based rank fraction" (Appendix C, p. 14) and "Effective rank is 23 of 64 dimensions," i.e., 23 of 64 (Appendix N, p. 22). Please add a Roy & Vetterli-style citation (Roy & Vetterli, "The effective rank: A measure of effective dimensionality," 2007) at first use so the entropy-based definition is properly attributed.
-
§3 — terminology for xi(j). The model section says "Let xi(j) denote recording j from participant i", but everywhere else (App. A; Fig. 2 caption) x∈R64 is the WavLM-PCA embedding of a recording, not the recording itself. Suggest: "Let xi(j)∈R64 denote the embedding of recording j from participant i."
-
§3 / §6.1 / App. M — the zero-mean assumption behind "averaging out". The argument that "recording-specific and stochastic variation can average out" for the attacker implicitly requires E[δij]=0 across recordings, but this is never stated in §3, §6.1, or App. M. App. E notes that background/vehicle acoustics are not explicitly removed — a plausible systematic per-participant bias that would violate zero-mean. Since the model is explicitly framed as explanatory, this is an implicit-assumption gap rather than an error; one sentence stating E[δij]=0 (and cross-referencing the App. E caveat) closes it.
-
Eq. (4) (§4) vs. Eq. (11) (App. A) — which content variable is released. Eq. (4) releases z=norm(D(c,s~)) written with c, while App. A and the prose state that the released representation is the posterior mean μc rather than a sample from q(c∣x). Suggest writing μc in Eq. (4), or adding a parenthetical that c denotes the posterior mean at release time.
-
§4 — bottleneck phrasing. "We apply a variational bottleneck to c" slightly outruns what governs the release: the information constraint acts on the stochastic q(c∣x), while the released map is deterministic in μc. App. A discloses this verbatim ("Stochasticity in the content branch is used only by the bottleneck objective"), so this is not misrepresentation — but §4 should mirror the disclosure with one clause, e.g., "...a variational bottleneck on the stochastic encoder q(c∣x); the released representation is the deterministic posterior mean."
-
Eq. (5) / App. G–H — standard deviation at K=1. The set summary concatenates the mean and standard deviation, but for a single-element set (K=1) the unbiased standard deviation is undefined and the population standard deviation is 0. The ddof convention is never documented, and every K=1 result (e.g., 0.528 in Table 1) depends on it. One sentence in App. G/H stating the convention (population std, or unbiased with a defined K=1 fallback) suffices.
-
Fig. 4(a) caption — undefined DR and colliding α^. The caption's "predicted DR∝R−1 scaling" uses DR, which appears nowhere else in the paper and is never defined, and reports a fitted exponent α^=−1.03 that collides with the dual variable α of Eq. (20). Suggest defining DR (e.g., the attainable disclosure at release count R) and renaming one of the two symbols — e.g., β^=−1.03 for the fitted exponent.
-
Fig. 4(b) / App. M — missing prediction formula. The "variance-based prediction of attainable identity signal" plotted in Fig. 4(b) is never written down, despite §6.1's explicit promise that "full derivations and variance-decomposition details are provided in Appendix M" — App. M currently contains only the 1/R identity, the dispersion series, and the variance shares. As a result, the claimed r=0.99 fit is unverifiable from the text. Please add the prediction formula to App. M.
-
App. M — variance-decomposition estimator. The estimator behind the 82.9%/14.8%/2.3% variance shares (Table 6) is unspecified — ANOVA-style on the random-effects model? moment-based? This is the same unfulfilled App. M promise. Please state the estimator explicitly (a sentence or two, or a citation).
-
App. M — R2 vs. R typesetting. In the sentence "The 1/R prediction gives R2 >0.9999 in each fold", the coefficient of determination is typeset identically to the release count R inside a sentence that is itself about R — genuinely confusing on first read. Suggest a proper superscript, R2>0.9999, or writing "coefficient of determination".
-
App. C, Eqs. (12)–(13) — pooling symbols and BCE variant. The bar-pooling symbols cˉ, zˉ(1), zˉ(2) are never bound to the prose pooling definition ("Participant pooling concatenates the mean and standard deviation of that participant's rows"), and whether the BCE in Eq. (12) operates on logits is unstated — inferable only from the sigmoid in Eq. (13). Suggest binding each symbol explicitly (e.g., cˉ=concat(mean,std) over the participant's rows) and stating which BCE variant is used (with-logits vs. on probabilities).
-
App. C, Eqs. (17)–(18) — undefined indices and aggregates. The only definitions given are "Let u+ and u− denote positive and negative linkage logits and let τ=0.5". Left undefined are: the indices i,j on ui+,uj−; the aggregation over (i,j) pairs (mean? sum?); the two-argument Huber0.5(⋅,0); and uˉ+, uˉ−, uˉ — including which logits enter uˉ. One defining sentence each would make these equations self-contained.
-
App. C, Eq. (19) — eigenvalue notation and rank-fraction formula. λi is never introduced as the eigenvalues of the class-conditional covariance Σ, and it collides with the loss weight λspk in Eq. (7). The effective-rank penalty (weight 4.0) has no formula — only the prose "activated when the entropy-based rank fraction falls below 0.3". Suggest renaming the eigenvalues (e.g., νi), adding a defining clause ("let νi denote the eigenvalues of Σ"), and giving the rank-fraction formula explicitly, e.g., a normalized entropy of the spectrum such as
or whichever form matches the implementation.
Thank you for the thoughtful review. We revised Sections 3-4, the Discussion, and the appendices to clarify the synthetic replacement mechanism, attacker design, utility constraint, and limitations.
W1. Synthetic identity sampling. MoIRAGE learns a 32-dimensional class-conditional diagonal Gaussian prior jointly with the encoder and decoder. At release, the source identity component is discarded and replaced by a sample combined with the deterministic content posterior mean. No diagnosis label is required at inference: the conditioning class is sampled from TRAIN class frequencies. Standard-normal and unconditional prior substitutions yield similar results, so our contribution does not depend on the specific prior distribution.
W2. DeepSets attacker. We clarified Eq. (5). A shared encoder processes each embedding; mean and population-standard-deviation pooling produces a permutation-invariant set representation (standard deviation is zero for K=1). A set-level network and symmetric head predict whether two collections belong to one participant. DeepSets is used during training, while independent attacker families are used for evaluation (Sec. 4; App. H).
W3. Clinical utility constraint. Clinical prediction, teacher distillation, and release consistency preserve task information. Separately, the dual constraint penalizes clinical loss exceeding 105% of its warm-up reference, discouraging degradation during privacy optimization. Appendix C specifies the complete objective, weights, training schedule, and dual updates.
W4. TPR/FPR. We now define True Positive Rate (TPR) and False Positive Rate (FPR) at first use. Under corrected size-controlled evaluation at K=10, TPR at 1% FPR falls from 0.738 to 0.018, and rank-1 identification from 0.876 to 0.087 (Table 1; App. L).
Q1. Extension beyond speech. Factorization, synthetic replacement, and collection-level adversarial training operate on embeddings and could extend to other longitudinal modalities. However, we validate only speech-derived representations, not imaging or wearables. The Bridge2AI variant removes ECAPA supervision but remains speech-based (Sec. 6.4).
Q2. Realistic attack assumptions. We assume attackers can access released embeddings, know the release procedure, and train with separate labeled reference participants. They do not need a target's raw audio or diagnosis. Our task is linking collections to the same participant, not recovering a legal identity. We clarify that this protection complements consent, access controls, data-use agreements, and institutional governance (Secs. 3, 7).
Q3. Anonymity and random replacement. Clinical content can retain identity even after replacement. The content-only ablation remains highly linkable (AUC⋆=0.982 at K=20), so replacing one factor does not establish anonymity. At K=20, Gaussian noise (σ=1) achieves linkage 0.622 but reduces clinical AUROC to 0.473. MoIRAGE achieves linkage 0.596 with AUROC 0.643. Our contribution is a better privacy-utility trade-off under repeated observations, not randomness alone. We claim neither anonymity nor differential privacy; formal guarantees require a defined privacy relation, bounded sensitivity, and repeated-release accounting (Secs. 6–7; App. K).
Discussion, limitations, and correction. We now acknowledge residual linkage, smaller eligible cohorts at larger K, adaptive attackers, and limited clinical-task coverage. Bridge2AI is a development replication, with lower AUROC for 10 of 12 auxiliary conditions. As explained in our common response, we found a collection-size cue in the submitted evaluation and replaced the primary results with exact-K, size-controlled evaluation of frozen release models and retrained attackers. Historical results are separately labeled.
We appreciate the reviewer's suggestions, which strengthened the clarity and scope of our claims.