Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2306.10269

Two simultaneous talkers distract more than one in simulated multi-talker environments, regardless of overall sound levels typical of open-plan offices

Manuj Yadav·Densil Cabrera
View original source ↗Read source document ↗Export metadata ↓

Abstract

The irrelevant speech effect (ISE) characterizes detriment to cognitive task performance in the presence of irrelevant speech. This paper examines whether the ISE varies due to the number of simultaneously active nearby talkers (for up to two talkers), or the overall sound level, within the context of a simulated open-plan office. Two experiments were conducted within a climate-controlled chamber that was set-up as a medium-sized open-plan office. The cognitive tasks performed by the participants included the digit recall task, and a writing task, within a room acoustic simulation of realistic multi-talker speech from spatially separated talkers. Within Experiment 1 (n=60), an increase in the number of talkers from none (T0) to one (T1), and from one to two (T2) simultaneous talkers resulted in statistically significant decline in the digit recall task performances, with effect sizes of 24% (i.e., T1 vs. T0), and 12% (i.e., T2 vs. T1), respectively. The pauses between words during the writing task were similar for T0 and T1, but showed a statistically significant increase within T2 vs. T1, with an effect size of 12%. The findings of Experiment 1 are inconsistent with the maximally distracting status attributed to T1 in some studies, but is consistent with findings in other studies. Within Experiment 2 (n = 62), the cognitive performance in T2 remained largely invariant between 45 and 57 dB (A-weighted sound pressure levels), which represents a typical range of levels within open-plan offices. In general, these findings have relevance for characterizing auditory distraction within complex multi-talker environments; both in laboratory studies and actual open-plan offices. (Abridged version; please see the paper for the full abstract)

Explore related subjects

eess.AScs.SD

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Manuj Yadav, Densil Cabrera. 2023-06-17. Two simultaneous talkers distract more than one in simulated multi-talker environments, regardless of overall sound levels typical of open-plan offices. https://doi.org/10.1016/j.apacoust.2018.12.007

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Recovering the Zipfian Distribution in Unsupervised Term Discovery

Unsupervised term discovery involves segmenting unlabelled speech into word- or syllable-like units and clustering these into a lexicon of candidate types. True lexicons follow a Zipfian distribution, yet the dominant centre-based clustering approach -- K-means -- produces a more uniform distribution due to an inductive bias toward spherical clusters. In this paper we revisit graph-based clustering as a bottom-up alternative, where segment embeddings are connected by pairwise similarity and partitioned using the Leiden algorithm. We show that graph clustering substantially outperforms centre-based approaches (K-means, GMM, BIRCH) in both word- and syllable-level lexicon discovery across three languages, producing more Zipf-like distributions. Another bottom-up approach, agglomerative clustering with average linkage, also performs well, although it is computationally less efficient and allows for less control over the resulting distribution. Our work calls into question the dominance of centre-based clustering for term discovery, and promotes graph clustering as an attractive alternative.

eess.AS↗
arXiv2026

Preserving Speech-to-Text LLM Capabilities in Speech-to-Speech Generation

Strong speech-to-text (S2T) LLMs already provide robust speech perception and text reasoning, but adding speech-to-speech (S2S) output is challenging: fine-tuning the backbone can degrade the original S2T performance, while attaching a downstream talker reintroduces a serial text-to-speech bottleneck. We present PRIME-Speech, a frozen-backbone S2S conversion framework that trains only speech-generation modules. PRIME-Speech synchronizes a causal audio post-decoder with intermediate hidden states of the frozen backbone, so codec tokens are generated from the model's evolving reasoning trajectory rather than from completed text chunks. The post-decoder uses mixed hidden-state, text, and audio-history conditioning, and a training-time packing strategy with turn-level audio KV-cache and position reset stabilizes multi-turn spoken interaction without additional multi-turn S2S training data. Multi-token prediction further reduces the effective codec prediction rate and improves first-audio latency without modifying the reasoning path. Across speech translation, spoken QA, speech understanding, and multi-turn dialogue, PRIME-Speech preserves the S2T behavior of the frozen backbone while producing accurate, low-WER spoken responses.

eess.AS↗
arXiv2026

Brain2Speech-Net: Fast and Intelligible Brain-to-Speech Synthesis Without Text Decoding

The loss of speech limits communication for individuals with paralysis. Direct neural-to-speech synthesis is challenging due to the limited availability of neural data for training speech brain-computer interfaces. Most existing systems rely on cascaded neural-to-text-to-speech pipelines, which increase inference latency and propagate errors across stages. We present Brain2Speech-Net, a single-stage neural-to-speech generation framework without intermediate text decoding. We use a differentiable phoneme bottleneck and a deep-HMM alignment mechanism to map long neural recordings into the latent space of a text-to-speech (TTS) model, enabling high-quality speech synthesis. Brain2Speech-Net is the only system in our comparison that produces intelligible speech while generating faster than real time.

eess.AS↗