Speech Language Models

Unlike autoregressive models, discrete diffusion-based models for zero-shot text-to-speech generate speech tokens in parallel and can revisit earlier predictions. Mask-and-replace training extends mask-only training by randomly replacing some tokens, and its gains are commonly attributed to self-correction, the ability to revise previously generated tokens. However, exposure to randomly perturbed context during training may itself improve generation, raising the question of whether these gains require inference-time token revision. To investigate this question, we use DeMaR, which combines mask-and-replace training with confidence-ranked mask-only sampling while preserving the total training corruption probability. Trained from scratch on LibriTTS, DeMaR achieves lower word error rates (WER) than autoregressive and mask-only diffusion baselines using the same speech tokenizer. This advantage persists when each token remains unchanged after first being unmasked. Matched training conditions on two heterogeneous speech tokenizers show that both noisy-context augmentation and replacement supervision improve WER under this restriction. These findings demonstrate training-side benefits of replacement beyond enabling inference-time token revision.

Speech Language Model (SLM) performance often degrades on dialects due to data scarcity. Conventional text-to-speech (TTS) augmentation struggles to cover diverse dialects as it requires a certain amount of real dialect speech. We propose synthesizing pseudo-dialect speech by converting LLM-generated dialect text via a standard-language TTS model, requiring zero real dialect speech. Additionally, we introduce intermediate standard-text prediction during training, acting as semantic normalization for downstream tasks. We evaluate dialect understanding via dialect-to-English speech translation across Japanese, German, and Chinese dialects. Compared to synthetic standard speech baselines, pseudo-dialect augmentation improves scores for Japanese (from 25.38 to 26.24) and German (from 31.57 to 32.47). Furthermore, the intermediate standard-text prediction effectively bridges the semantic gap, boosting performance to 28.26 for Japanese and from 11.67 to 16.37 for Chinese. These results suggest that our approach scales to various languages without requiring speech resources specific to each dialect.

Speech large language models (speech LLMs) perform well on automatic speech recognition (ASR) when sufficient paired speech-text data is available, but their performance degrades in low-resource settings. A cascaded pipeline that performs speech-to-phoneme (S2P) conversion followed by phoneme-to-grapheme (P2G) conversion has been shown to outperform end-to-end speech LLMs in this regime, suggesting that phoneme-mediated processing is beneficial when paired data is scarce. We propose phoneme-guided initialization, a simple method that uses this insight within an end-to-end framework: we pre-train the audio encoder on S2P and the LLM on P2G tasks, then connect them and fine-tune the full model end-to-end on the target ASR task. Experiments on Japanese (CSJ), Chinese (AISHELL-1), and two low-resource languages from Common Voice 25.0 (Tatar and Urdu) show that our method matches or outperforms both the cascaded S2P-P2G baseline and the end-to-end model without P2G initialization.

Automatic Speech Recognition (ASR) systems often show uneven performance across demographic groups, and errors can be especially difficult to address for speakers belonging to multiple demographic groups. This work studies demographic-aware model merging for fair Speech-LLM-based ASR. Starting from a SLAM-ASR-based model, we fine-tune only the connector on demographic-specific subsets and merge the resulting subgroup-adapted connectors into a global model. We then identify critical cross-axis demographic pairs using subgroup WER and task-vector conflict, and apply intersection-specific correction vectors to the global merged model. Experiments on Fair-Speech show that global demographic merging improves overall WER over the base model, while intersection correction provides additional gains for several merging strategies. In particular, TIES with WER-based correction achieves the best overall WER, reducing it from 7.38% to 5.13%. Subgroup and disparity analyses further show that the proposed approach improves performance across demographic axes, while highlighting that lower average WER does not always imply reduced subgroup disparity.

Full-duplex voice agents need to modulate emotion and delivery during real-time conversations, when de-escalating a complaint, carrying urgency in dispatch, softening a clinical result. Emotion and delivery control is well studied for TTS and turn based models through prompt-conditioned synthesis, reference-conditioned synthesis and activation steering; PersonaPlex controls identity in a duplex model but not affect. We study emotion steering in Moshi, a fully open sourced full-duplex speech language model, across four emotions, using mean-difference activation steering, which costs only a few vector additions per frame and no retraining. We show that emotion is linearly decodable from Moshi's residual stream, but activation steering is only partially achievable, and unevenly so; as happy, angry and surprise steer towards a shared direction while sad is distinctly steerable. We also show that the shared component across the three emotions cannot simply be projected away from all the emotions equally.

Speech-LLMs often exhibit prompt overfitting, where models solely trained on automatic speech recognition (ASR) instruction fail to generalize to new instructions such as speech translation and continue to behave primarily as ASR system. We propose DirectSpeech2LLM, a simple end-to-end framework that preserves the instruction-following ability of the LLM on unseen tasks when conditioned on speech. It computes distance-based CTC loss over the frozen LLM embedding matrix and uses greedy CTC labels to derive geometrically and temporally aligned speech embeddings respectively as an input to the LLM. Trained solely on 960 hours of LibriSpeech ASR data, DirectSpeech2LLM outperforms the cascaded system on ASR (seen task) and generalizes zero-shot to speech translation and emotion recognition (two unseen tasks), closely matching the cascaded system upper bound on these two new instructions despite seeing neither during training. We also find that geometric alignment strength plays a smaller role than previously assumed, as our modified CTC loss is shown to provide sufficient implicit geometric grounding without requiring an explicit regression loss. Results are consistent across two LLM families and scale with both more training data and model capacity.

This paper presents the HINTT system submitted to the 2nd Challenge and Workshop on Multilingual Conversational Speech Language Model (MLC-SLM). We address multilingual speaker-attributed ASR, where systems must determine who spoke when and what was spoken. We investigate two modeling strategies for this problem: a cascaded pipeline that combines speaker diarization with speech-LLM-based ASR, and a unified speech LLM that directly generates speaker labels, timestamps, and transcriptions. Our final submission is based on the cascaded pipeline, consisting of a fine-tuned DiariZen diarization model, a fine-tuned Qwen3-ASR model, and LLM-based generative error correction. For comparison, we also fine-tune VibeVoice-ASR as a unified model using the same official training data. All task-specific fine-tuning and model selection are performed using only the official MLC-SLM data, without external data or pseudo-labels. Experimental results demonstrate that the cascaded system remains more reliable under the MLC-SLM Task 1 conditions, while unified speech LLMs offer a promising direction for future speaker-attributed ASR.

As human--AI interactions become more conversational, full-duplex speech language models capable of natural real-time dialogue are growing in importance. Beyond generating appropriate responses, these models must coordinate turn-taking, backchanneling, and floor management in real time. Reinforcement learning (RL) provides a way to refine these behaviors through direct feedback on interaction outcomes. However, existing RL methods either apply timing feedback to a token policy or optimize semantic content, leaving the joint improvement of timing and content unresolved. We introduce HiPLEX, an RL framework that factorizes a pretrained full-duplex text policy into a control policy that decides when to emit content and a conditional content policy that decides what to emit. The first factor selects among 'pad', 'epad', and 'con'. The second selects a token only when 'con' is chosen. This hierarchy describes conditional actions within each frame and uses the model's existing text head. We route timing advantages to the token-group factor through event-causal masks derived from generated speech episodes, and route an LLM-judge semantic advantage to the conditional content factor. Across three Moshi seeds on Full-Duplex-Bench v1, HiPLEX reduces takeover rates during natural user pauses and backchannel opportunities, and shortens post-interruption response latency relative to GRPO, while maintaining comparable judged interruption-response quality. On Moshi and PersonaPlex, HiPLEX better matches pooled human turn-timing and backchannel-rate marginals than GRPO.

LLM-based text-to-speech (TTS) models can generate emotionally expressive speech, but how reference emotion is routed through the model and realized in decoded speech remains unclear. We introduce two emotion-sensitive metrics for matched neutral and emotional syntheses---a codec trajectory score and a late residual direction score---and use them to score activation-patching interventions. Under controlled matched-reference conditions, this analysis identifies a sparse source-to-readout component-level circuit: 23--27 attention heads and MLPs per emotion, roughly 5% of the components considered, recover or suppress 74--88% of the late emotion-readout shift on held-out cases. The circuit combines a shared component backbone with emotion-specific components; cross-emotion activation swaps reduce the target readout in 47 of 48 cases. In decoded speech, the same intervention produces consistent changes in pitch, energy, and spectral brightness over 24 matched pairs per emotion. A readout-matched residual-direction baseline produces only 17--27% of the intervention's pitch effect, showing that internal readout movement alone does not explain the decoded acoustic changes. These results trace a compact causal route from reference-derived prefix information to emotion-relevant properties of generated speech.

Empathetic spoken dialogue is a sophisticated cognitive process that requires not only recognizing emotions but also inferring a user's latent mental states to provide appropriate support. However, current SpeechLLMs often treat empathy as a direct input-to-response mapping, leading to "superficially warm" but emotionally hollow interactions. In addition, since empathy relies on a multi-stage process with strong inter-step dependency, errors at any intermediate step can cascade through subsequent steps and lead to inappropriate responses, while existing training paradigms lack mechanisms to precisely localize and improve such errors. In this work, we propose EchoChat, a unified framework that reformulates empathetic spoken dialogue as a structured cognitive reasoning process integrating perception, mental-state reasoning, and response generation. To support this paradigm, we first construct EchoDialogue-400K, an acoustically rich dataset for multi-stage empathetic supervision. During the SFT stage, we strengthen acoustic grounding through proposed Acoustic-Anchored Attention (AAA). During the RL stage, we further introduce a novel stage-aware optimization objective with Step-Decomposed Credit Assignment (SDCA) to localize reasoning errors and mitigate cascaded error propagation. In addition, we introduce EchoEval, an expert-annotated benchmark for multi-dimensional empathy evaluation. Extensive experiments demonstrate that EchoChat achieves state-of-the-art performance in perception, reasoning, and response alignment. Project page: https://github.com/dingdongwang/EchoChat

Activation steering has proven effective for controlling the behavior of Large Language Models (LLMs) at inference time, but its application to SpeechLLMs remains new, and training-free steering approaches for such models are still largely unexplored. We propose a training-free Contrastive Activation Addition (CAA) protocol that derives steering vectors for common speech tasks (e.g. transcription) in SpeechLLMs from a small number of labeled utterances. We showcase that adding these vectors in the representation space, at inference time, enforces better the targeted speech task. We further show that, when combined with prompting, these vectors yield to consistent improvement over prompting alone on most evaluated tasks such as Automatic Speech Recognition (ASR) or Emotion Recognition (ER), and transfer to out-of-domain data. We additionally demonstrate the usefulness of script-normalization directions to enforce the target script of a specific language.

Large speech language models have demonstrated strong capabilities in unified cross-modal understanding and generation, yet paralinguistic cues, especially emotion, remain difficult to preserve. Existing systems typically rely on entangled acoustic representations, which allow the underlying language model to depend excessively on recovered lexical content instead of grounding its behavior in acoustic-prosodic evidence. We address this limitation with EMODE, an emotion-aware speech language model built around Dynamic Para-Semantic Experts (DPSE). DPSE decomposes continuous speech features into semantic and paralinguistic pathways, routes them dynamically, and fuses them before integration into the language model. To turn this structural decomposition into functional specialization, EMODE is trained with a three-stage curriculum consisting of semantic warm-up, paralinguistic activation, and joint refinement, guided by Orthogonal Expert Guidance (OEG), Semantic-to-Acoustic Alignment (SAA), and Gating Diversity Regularization (GDR). Experiments on SER test, empathetic response evaluation, and the newly constructed bilingual MEPA benchmark show that EMODE improves the balance between lexical fidelity and emotional sensitivity, strengthens affect-grounded response generation, and exposes the value of explicit para-semantic factorization for robust cross-corpus emotion understanding.

Delayed Streams Modeling (DSM) enables LLM-based streaming automatic speech recognition (ASR) by aligning acoustic and text streams on a common timeline. DSM adds the padding token <p> and the word-start token <w> to the LLM vocabulary and predicts them together with normal text tokens using the same softmax. We first show that <w> can be removed while maintaining competitive recognition performance. Based on this result, we propose Factorized DSM (F-DSM), which separates the waiting probability for <p> from the distribution over the original LLM vocabulary. This factorization removes ASR-specific tokens from the text prediction space and allows the large-vocabulary softmax to be skipped on waiting steps. Experiments on the Corpus of Spontaneous Japanese and LibriSpeech show that F-DSM achieves better recognition performance than DSM. It also greatly reduces GPU memory use while maintaining similar training throughput, provides a small inference speed improvement through softmax skipping, and reduces the degradation in text-only perplexity observed with DSM.

Speech delivery varies with both the requested paralinguistic attribute and the linguistic content. We introduce ParaGeo, a matched-content decomposition of paralinguistic variation in a frozen speech language model. Synthesized audio tokens are replayed with a fixed listening prompt; pooled key/value (K/V) representations are centered and projected into a shared low-dimensional space. Our GLM-4-Voice probe spans 80 requested controls from 12 benchmark families across eight sentences. With a globally fitted calibration basis, content-held-out centroid accuracy using this basis is 9.49% versus a 1.25% permutation baseline; same-label cross-content cosine similarity is 0.285 versus 0.017, and both conditional permutation tests yield p = 0.001. A separate ten-scenario, six-style probe reveals reproducible contrast directions across scenarios. Static, additive, and temporal interventions produce attribute-, layer-, and schedule-dependent response profiles. These results provide a shared coordinate representation for measuring paralinguistic structure and an empirical starting point for latent speech control. Code is available at https://github.com/yuhanlydia/ParaGeo.

Full-duplex voice agents make many small, closed decisions, which current systems answer by slow autoregressive decoding. We propose DuplexJev, which feeds ASR-encoder hidden states through a small connector into a frozen LLM and reads each question as a single-token distribution over its options. Nothing is decoded, and an 8-GPU node answers 80 decisions about eight utterances in about 0.1 s. With a last-layer connector, spoken QA stays close to reading the transcript (90% vs. 91%). DuplexJev also hears the speaker: gender and emotion accuracy both reach 90% (from 55% and 28%) with a cross-attention connector, whose spoken QA drops by only 1 point (83% to 82%). We train decisions with cross-entropy on the read-out answer token, instead of the usual transcript distillation, whose teacher never hears the voice, and keep distillation for content. Encoders and LLMs are interchangeable; we release weights, training recipe, a batched-inference pipeline for full-duplex serving and a bilingual spoken-QA set.

Simultaneous speech translation must emit useful target text before the source is complete while preserving every committed token. We adapt a full-utterance speech language model using prefix supervision derived from its own complete- and partial-waveform translations, requiring neither transcripts nor human translations. We compare single-turn forced-prefix and multi-turn append-only decoding, use a confidence threshold to control the inference-time quality--latency trade-off, and vary the density of training prefixes with a separate synthesis margin. On FLEURS and CoVoST2 in three language directions, prefix training improves quality--latency frontiers over the unadapted model, and confidence provides the broadest consistently competitive operating range. Multi-turn decoding is generally stronger at low latency; under multi-turn training, commit-calibration error falls by 63--68% overall and 68--80% at early prefixes, whereas single-turn training provides only modest overall calibration gains and no early-prefix improvement. A small synthesis margin sometimes extends the frontier to lower latency, particularly on shorter utterances, while a larger margin degrades translation quality and calibration. Prefix adaptation therefore improves simultaneous speech translation, especially under multi-turn append-only decoding, while synthesis density introduces a non-monotonic quality--latency trade-off.

Model intelligence and fast response jointly shape the quality of interaction with speech language models, yet remain difficult to achieve together. Explicit chain-of-thought (CoT) improves reasoning and audio understanding, but generating intermediate reasoning tokens delays responses. Describing fine-grained acoustic cues further lengthens CoT and increases latency. Latent reasoning can reduce this overhead, yet existing methods often trail CoT and remain limited by single-path supervision and reasoning budgets that do not adapt to problem difficulty. We introduce AURAL, which models a distribution over multiple plausible reasoning continuations in latent space and jointly predicts chunks of future states to reduce sequential forward passes and reasoning latency. To provide initial supervision for latent reasoning, we construct AuralReason-683K: 683K bilingual speech utterances (about 1,000 hours) with concise CoT for emotion recognition, empathetic dialogue, and general reasoning. AURAL-RL then explores beyond these traces, rewarding concise reasoning that yields high-quality answers and adapting reasoning effort to each problem. Across two backbones, AURAL-RL achieves performance comparable to CoT-RL, with larger gains over the respective supervised checkpoints on most metrics. Analysis further shows that harder questions elicit more latent reasoning steps. On Qwen2.5-Omni, it reduces time to the first answer token by 11.8x, from 1.22 to 0.10 s, versus 0.05 s for direct answering.

Neural speech codecs increasingly serve as tokenizers for speech language models (SLMs). Lowering the frame rate reduces the computational and memory costs of SLMs, but makes it difficult to preserve both linguistic information and acoustic detail. Existing approaches rely on rule-based compression: average pooling can discard linguistic information, whereas similarity-based merging uses a fixed threshold on adjacent-frame similarity and applies the resulting boundaries to the acoustic stream. We propose Q-SPT, a low-frame-rate dual-stream speech tokenizer with separate, context-aware, learnable query-based compressors specialized for semantic and acoustic representations. In particular, queries at a fixed rate independently attend to the semantic and acoustic streams as separate key-value sources, enabling stream-specific, context-aware aggregation through two separately learned compressors. In addition, an autoregressive text loss explicitly supervises the semantic compressor to preserve linguistic information. Experimental results show that Q-SPT achieves the best reconstruction among the evaluated codecs at the same frame rate. In downstream SLMs, it yields the best speech recognition accuracy and text-to-speech perceptual quality with competitive intelligibility.

Speech emotion recognition (SER) is the task of assigning emotion labels to utterances. Early systems relied on acoustic features, whereas recent approaches combine multiple modalities, most commonly speech and text. Still, performance remains poor on many datasets. Large language models (LLMs) have therefore attracted interest for SER, as they can process diverse inputs jointly with instructions. However, direct audio input raises questions of explainability. To address similar questions in image classification, concept bottleneck models were introduced. This work adapts concept bottlenecks to SER to examine how individual predictions depend on transcripts, acoustic descriptions and speaker attributes. Experiments test three LLMs on CREMA-D, IEMOCAP and MELD, with concepts extracted by separate tools. On scripted corpora, LLMs are strongly biased towards the transcript in the zero-shot setting, which lowers Macro-F1 from 27.8 to 5.8 on CREMA-D. Fine-tuning removes this bias, and the transcript raises Macro-F1 from 41.8 to 45.1. Removing speech rate changes 48% of Neutral predictions to Disgust on CREMA-D; removing intensity level on MELD changes predictions despite little change in Macro-F1. These findings show that aggregate performance changes alone do not capture the effects of concept removal on individual predictions.

Demographic fairness gaps in automatic speech recognition are almost always reported from a single training run. We fine-tune the Q-former projector and LoRA adapters of a speech LLM at five audio compression factors and six random seeds, holding the encoder, base decoder, data and decoding fixed, and evaluate every run on Common Voice and Fair-Speech. At 460 h of clean LibriSpeech, the seed moves fairness metrics more than compression does on most demographic axes. A balanced 3x3 decomposition attributes 85.3% of the variation in Fair-Speech ethnicity normalized gap to the seed against 8.3% to compression (p = 0.009), though compression explains more on age and gender. Held-out LibriSpeech word error rate spreads by 0.04 points across those seeds while Common Voice spreads by 8.57, so these are not failed runs, and the effect survives controlling for accuracy and dropout. Scaling and diversifying the adaptation set to 960 h damps the effect but does not remove it. On Fair-Speech ethnicity, two single-run systems must differ by more than 0.30 in normalized gap to exceed seed variability.

Speech-LLMs are expensive to run, making compression important for real-world deployment. However, compressed models are usually selected using aggregate word error rate (WER), which can hide how pruning affects different demographic groups. In this work, we systematically study the effect of audio encoder pruning on SLAM-ASR for different demographic groups. Using the Fair-Speech and Common Voice datasets, we found that the pruning does not affect all demographic groups equally; the gap between best- and worst-performing groups increases in fold. These disparities appear across all three encoder scales, but only the largest model initially hides them behind aggregate WER. LoRA adaptation improves WER for every group, but benefits groups already performing well more strongly and widens for certain groups. On Common Voice English, Danish, and Dutch, accent gaps persist but do not clearly widen, showing that the fairness effects of pruning vary across datasets and must be measured directly. Our findings suggest that for pruned models, deployment decisions should include per-group WER, with the worst-performing group's error rate as an explicit criterion.

Empathetic spoken dialogue requires models to use both what is said and how it is said to decide how to respond. Explicit CoT can improve paralinguistic perception and make acoustic cues more explicit in replies, yet does not ensure their effective use in response planning. We call this mismatch the perception-reasoning gap. In addition, CoT may not fully capture acoustic cues in words, and generating it adds inference latency. To address these limitations, we introduce LoopSLM, which builds on looped Transformers for latent reasoning, reusing a decoder block to refine hidden states with acoustic grounding at every pass. Its two-stage training further narrows the perception-reasoning gap by separating learning to reason from learning to respond, enabling direct inference without CoT. On EchoMind, LoopSLM improves paralinguistic understanding, reasoning, and reply quality over Qwen2.5-Omni-7B. Against the CoT-SFT baseline, LoopSLM gains over 20 points in reasoning accuracy while generating 64.5% fewer tokens at half the latency. It also outperforms Qwen3-Omni-Thinking on most empathetic reply metrics with 34x lower latency. Despite training only on dialogue data, LoopSLM improves accuracy on general audio benchmarks.

We present JoyAI-Voice~2.0, an end-to-end anthropomorphic speech generation model built upon a fully continuous, dual-encoder architecture. Raw speech is encoded into continuous latents and partitioned into patches. Each patch is decomposed by a semantic-acoustic dual encoder into a semantically purified representation and an acoustic representation, which are fused and jointly fed to a causal autoregressive Transformer for planning. The Transformer predicts the conditioning for the next patch, and a local diffusion Transformer renders its full latents for 48,kHz synthesis. The model is trained with a joint flow-matching and stop-prediction objective, followed by supervised fine-tuning and reinforcement learning with DiffusionNFT to improve model performance. It achieves the lowest average word error rate of 2.51% on Seed-TTS, a 14.9% relative reduction over the strongest baseline, and state-of-the-art attribute fidelity on InstructTTSEval in both Chinese and English, leading on 5 of 10 perceptual dimensions with the highest overall score of 0.893 on MDVD-Eval.

LLM-as-a-judge is widely used for evaluating text, but extending this paradigm to speech requires models to interpret acoustic as well as linguistic evidence. This introduces a modality-specific risk: a speech judge may treat a perceptually salient cue as evidence of quality even when that cue is irrelevant to the target criterion or receives more weight than human listeners give it. We call this behavior an acoustic shortcut. To study it, we audit six speech LLM judges using controlled manipulations of intensity, content richness, and emotional delivery. We evaluate both pointwise scoring and pairwise comparison, using human preference calibration to interpret the results. The judges consistently reward louder audio, prefer content-rich speech more strongly than human listeners do, and map emotional delivery into quality preferences. These effects are most visible in pairwise comparison, while pointwise scores often obscure them. More concerningly, the accompanying rationales rarely identify the acoustic cue that changes a judgment and instead repeatedly rely on a limited vocabulary, leaving them acoustically underspecified. Together, these findings show that reliable speech judges must both resist acoustic shortcuts and ground their rationales in the acoustic evidence behind their decisions. To support reproducibility and future audits, we also release SpeechJudgeAudit, the controlled stimuli and evaluation tools used in this study.

End-to-end full-duplex speech models have brought open-source machine conversation closer to human-like interaction, yet existing systems remain limited in two intertwined dimensions: long-context robustness and multi-party interaction. Real-world scenarios such as meetings, group lessons, and social-robot reception require a single model to track, contextualize, and respond to multiple speakers over extended durations. Progress is constrained by both data and evaluation: open multi-party speech corpora remain small and are not designed for codec-frame-level full-duplex modeling, while existing long-audio benchmarks focus on passive listening and speech-to-speech benchmarks are mostly short and dyadic. We extend the Moshi paradigm jointly along the long-horizon and multi-party axes in English and Chinese. First, we release 57.6k hours of synthetic training data ($[MultiTalkPT](https://huggingface.co/datasets/MultiTalk/MultiTalkPT)$ and MultiTalkFT) for long-form, multi-party, English-Chinese full-duplex dialogue, with controllable length, participant count, turn-taking, overlap, backchannels, interruptions, addressee shifts, and long-range coreference. Second, we introduce MultiTalkBench, built from real human recordings, for evaluating long-form, multi-party, bilingual full-duplex dialogue. Conversations average 32.6 minutes and include probes for long-range entity tracking, topic coherence, and addressee selection. Third, we train a bilingual Moshi-style model that sustains coherent multi-party English-Chinese conversations over extended durations and substantially outperforms open-source baselines including Moshi, MiniCPM-o-4.5, and Qwen3-Omni-30B-A3B-Instruct on MultiTalkBench.

Streaming spoken agents may produce the correct final action after acting too early. Final-turn scores do not reveal whether each observed speech prefix supports an exposed action. We introduce the Partial Speech Action Contract for Turn Taking in Speech Language Models (PACT-SLM), a controlled test that assigns the first valid action time and evaluates both action identity and timing. In the primary test, 80 paired contrast groups from four held-out semantic families yield 1,600 prefix predictions across clean and 15 dB noise renderings. Using source-utterance semantic targets rather than counterbalanced branch codes, WavLM Base Plus reaches 26.03% pooled post-onset semantic-label accuracy (95% group-bootstrap interval [22.14%, 29.68%]), exposes an action on 18.99% of pre-onset prefixes, and predicts 5.94% of complete trajectories exactly. Its pooled label score is at the 96th percentile of 100 within-prefix label permutations, below the 97.5th-percentile reference (26.73%). It exceeds matched text, scalar-acoustic, and shuffled-representation probes in pooled post-onset label accuracy. Elapsed time is more onset-exact (36.25% versus 23.13%) but less accurate about action identity (9.92% versus 26.03%). These results motivate separate measurement of action identity and onset timing in partial-speech evaluations.

Human speech conveys rich perceptual information, such as emotion and speaker identity, yet most automatic speech quality judges reduce it to a single naturalness score. We study diagnostic speech judges: given two candidates, a diagnostic judge decides which is better, along which perceptual dimensions (e.g., timbre, emotion, timing) they differ, and which audible cues support its decision. Learning such judges is challenging: expert annotation is costly, and simply prompting a frontier audio-language model to produce labels is unreliable: our probing reveals substantial errors and unstable instruction following. We introduce SpeechCritic, which learns a diagnostic judge in a reference-conditioned cross-lingual setting from only about 300 human-labeled comparisons. Rather than replacing the frontier model, SpeechCritic calibrates it with these labels: for each dimension, it selects the acoustic measurements that agree with human judgments, maps them to A/Tie/B probabilities, and passes these to the model as non-binding hints alongside the audio. Compared with the same model labeling without hints, this raises dimension-level agreement with humans by 6.3 points and cuts the mismatch with human Tie rates by 10.4 points. We then train a 7B judge on this supervision and find that different training signals shape different judge behaviors: SFT establishes the task, OPD transfers the teacher's dimension-level strengths and weaknesses, and RL helps most on clear-cut comparisons where human raters agree. Notably, human listeners also find that RL makes rationales cite more specific, localized acoustic cues, although it never directly rewards rationale text. Finally, we show that the pipeline is language-pair agnostic by instantiating it on both English-Japanese and English-Spanish. Together, these results demonstrate a path from limited human preferences to a diagnostic speech judge.

Alzheimer's disease (AD) and mild cognitive impairment (MCI), which may precede AD, manifest early through subtle linguistic and acoustic alterations. Traditional diagnostics, however, are often resource-intensive and lack scalability for mass screening. To address these challenges, we introduce a novel bilingual speech large language model framework for automated, explainable cognitive screening. Unlike conventional pipelines that rely on error-prone automatic speech recognition, our system directly processes raw speech to learn joint acoustic-semantic representations, preserving critical prosodic cues often lost in transcription. Utilising our newly collected PUTH-AD dataset alongside multiple open-source corpora, we implemented a multi-task learning objective that simultaneously performs cognitive status classification and generates clinician-understandable natural language explanations. Our system achieved the highest average accuracy and AUROC across six dataset/task conditions, comparing three representative baselines. The system demonstrated cross-task transfer to held-out PUTH-AD task subsets, maintaining classification accuracy on an entirely unseen cognitive task without task-specific fine-tuning. Furthermore, clinician evaluation confirms that the generated explanations are both clinically relevant and largely consistent with the underlying speech evidence, supporting their potential utility in clinical interpretation. This study provides a scalable, objective, and explainable framework for speech-based cognitive screening, combining cognitive status classification with natural language explanations that clinicians can assess and verify, bridging the gap between advanced AI and clinical utility.

Speech-to-speech translation (S2ST) in time-sensitive applications such as video dubbing requires not only semantic fidelity and speaker preservation, but also strict duration consistency to avoid audio-visual misalignment. However, existing S2ST systems largely generate target speech without explicit temporal planning, making duration control an unresolved challenge. We introduce DuraS2ST, a duration-aligned reasoning framework that enables a single speech language model to first generate an explicit chain-of-thought (CoT) for planning target wording and phonetic length, and then synthesize the corresponding speech tokens. To support this paradigm, we construct DuraSet-440K, a high-quality duration-aligned CoT corpus for supervised initialization. We further optimize the model with multi-modal multi-dimensional reinforcement learning, using a Duration Margin Reward to balance translation quality and duration consistency, and Modality-Aware Reward Attribution to assign rewards to appropriate token spans. Experiments on CVSS-T show that DuraS2ST achieves a strong balance between translation quality and duration consistency, outperforming competitive open-source and commercial baselines. Project page: https://github.com/Mia11939/DuraS2ST.

Full-duplex spoken dialogue models can listen and speak simultaneously like the real-time dynamics of human conversation. For natural dialogue, the ability to search for external information in real-time is also an important capability. Many models remain trapped in parametric knowledge, leaving them unable to access real-time information and tool execution. Furthermore, even when Large Language Models (LLM) retrieve information, many duplex speech models process it within a compressed latent space rather than in its raw text form, which can lead to information loss from compression. To address this issue, we propose Context Spanning, a framework for information injection between a full-duplex speech model and an external LLM backend via real-time chunked prefill. The injected frame is encoded in a single forward pass inside the real-time frame budget. It feeds the retrieved information to the speech model as-is, enabling it to reason over the information independently and generate responses. With this approach, our model achieves high performance on Full-Duplex benchmarks and strong results on Question Answering tasks, demonstrating its conversation potential. Context Spanning shows that external information can be injected directly into a duplex speech model, introducing a new simple and powerful mechanism for duplex systems.