# Batched Speech Decisions Without Decoding: Single-Token
Source: https://savrn.com/papers/batched-speech-decisions-without-decoding-single-token-supervision-lets-a-frozen-llm-hear
Markdown alternate of the page above; the site index is https://savrn.com/llms.txt

---

## Abstract

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.

[Full paper on arXiv](https://arxiv.org/abs/2610.02638)

## Details

arXiv identifier

2610.02638

Published

2026-10-02

Authors

Jie Jin, Ziyin Ma, Min Yin, Jinyu Chen, Haigang Song, Zhikun Pang, Xiaowen Zhang

## Open Models Built on This Paper

Every model in the SAVRN Model Hub whose card cites this paper, most downloaded first, with what it takes to run each one.

| Model | Task | Size | License | Monthly downloads | Cheapest setup at 16-bit |
| --- | --- | --- | --- | --- | --- |
| [DuplexJev-32B-Stress](https://savrn.com/models/duplexjev-32b-stress) Adventists.ai | [Audio text to text](https://savrn.com/models/tasks/audio-text-to-text) | — | cc-by-nc-4.0 | — | — |
| [DuplexJev-4B-Stress](https://savrn.com/models/duplexjev-4b-stress) Adventists.ai | [Audio text to text](https://savrn.com/models/tasks/audio-text-to-text) | 4.2B | cc-by-nc-4.0 | — | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) $1.85/hr |
| [DuplexJev-32B-Para](https://savrn.com/models/duplexjev-32b-para) Adventists.ai | [Audio text to text](https://savrn.com/models/tasks/audio-text-to-text) | — | cc-by-nc-4.0 | — | — |
| [DuplexJev-4B-Para-LiveTurn](https://savrn.com/models/duplexjev-4b-para-liveturn) Adventists.ai | [Audio text to text](https://savrn.com/models/tasks/audio-text-to-text) | 4.2B | cc-by-nc-4.0 | — | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) $1.85/hr |

## Explore More

- [All research papers](https://savrn.com/papers)
- [The model directory](https://savrn.com/models)
- [Datasets](https://savrn.com/datasets)

## Source

- arXiv metadata, read 2026-10-14.
- [How the hub is built](https://savrn.com/model-hub/methodology)