arXiv:2608. 08569v1 Announce Type: new Abstract: Recent advancements in Speech Large Language Models have demonstrated remarkable capabilities in understanding complex audio tasks.
VoiceLongMemEval (VLME) is a new benchmark that tests AI assistants on their ability to remember how users sounded by incorporating paralinguistic metadata—such as emotion labels, prosody descriptors, and voice events—into each conversational turn. The benchmark uses a three‑stage adversarial gate to ensure that models cannot succeed with transcript alone, revealing a significant affect gap: models gain 0.09 to 0.38 accuracy when provided with paralinguistic cues, and audio‑native models outperform standard ASR pipelines in extracting these signals. The dataset and code will be released upon acceptance.
DiaLong is a new long‑context dialogue benchmark that tests transformer‑based Large Language Models’ ability to retain and recall memories. The dataset contains 100 human‑created conversations, expanded with LLMs to exceed lengths of existing datasets, and includes a fact‑recall task where models must classify 40 candidate facts as true or false. Experiments show current LLMs and memory‑enhancement methods still struggle, indicating a need for improved memory performance in extended conversations.
SONIC‑O1 is a new benchmark designed to evaluate multimodal large language models on audio‑video understanding. It contains 60 hours of 231 clips across 13 real‑world conversational domains, with 4,958 human‑verified annotations and demographic metadata. The benchmark tests open‑ended summarization, multiple‑choice question answering, and temporally grounded reasoning, revealing performance gaps between model families and across demographic groups.
arXiv:2609.23416v1 Announce Type: cross Abstract: Long-form audio performance is often summarized by context length and aggregate accuracy, obscuring how language, evidence, and task jointly shape di...
arXiv:2609.22697v1 Announce Type: new Abstract: Recently, text-to-speech systems have made significant progress in speech expressiveness and controllability. However, the speaking style of generated...
The Eloquence team presents three methods for the Interspeech 2026 MLC‑SLM Task 2, a multilingual MCQA challenge covering 21 languages. They fine‑tune Voxtral‑Mini‑3B with LoRA and data augmentation, achieving 0.72 macro‑accuracy; they use multimodal in‑context learning on Voxtral‑24B to correct label bias, reaching 0.81; and they deploy a training‑free retrieval system with a voice‑anchored memory, scoring 0.68. All approaches surpass the official baseline.
arXiv:2609.22214v1 Announce Type: new Abstract: Long multilingual conversational spoken question answering requires systems to balance long-range transcript semantics with sparse acoustic and speaker...
arXiv:2607. 05365v1 Announce Type: cross Abstract: Streaming speech-to-speech language models aim to answer spoken queries directly with synthetic speech.
The paper introduces MAD2, a synthetic benchmark of 1,000 two‑speaker dialogues with about 10 hours of audio and 1,230 check‑worthy sentence annotations for spoken claim verification. It proposes a calibrated multimodal fusion approach that combines a context‑aware audio encoder with a dialogue‑aware text model. Experiments show that adding dialogue context improves verification performance, though the gains differ across scenarios, and that fusion offers the largest advantage when full‑dialogue context is available, though it does not consistently outperform text alone.
The Eloquence team presents three strategies for the Interspeech 2026 MLC‑SLM Task 2, a multilingual MCQA challenge across 21 languages. They fine‑tune Voxtral‑Mini‑3B with LoRA and cross‑lingual augmentations, achieving 0.72 macro‑accuracy; they use multimodal in‑context learning on the frozen Voxtral‑24B to correct label bias, reaching 0.81; and they deploy a training‑free retrieval system with a voice‑anchored memory, scoring 0.68. All approaches surpass the official baseline.
Streaming speech-to-speech language models aim to answer spoken queries directly with synthetic speech. However, standard speech and text benchmarks do not capture whether these systems behave naturally in conversations, where timing, turn-taking, prosody, interpersonal stance, language and dialect consistency, and relationship-aware appropriateness jointly shape perceived quality.