arXiv cs.CL· Solano Amadori·· 2 天前AI 评分31
Marimba:基于冻结 Qwen3-ASR 0.6B 特征的语音决策系统模型
Marimba: A System One Model for Speech
AI 导读
Marimba 是一个基于冻结 Qwen3-ASR 0.6B 特征的语音决策接口,可判断音频窗口内的词项存在、话轮结束和语种。在私有对话测试集上解码器 AUROC 为 0.80-0.92,公开朗读语音上为 0.92-0.97;Speech Commands 四分类准确率 0.93。
正文 · 原文
Abstract:Speech applications require decisions about term presence, turn completion, and language within an audio window. We present Marimba, an interface for typed speech decisions built on frozen Qwen3-ASR 0.6B features. Lexical questions use zero-shot scoring of ASR transcripts; non-lexical questions use a question-conditioned head and a separate layer-18 turn-end readout. Language identification uses the ASR model's own language scores. On private conversational test sets, decoder AUROC is 0.80-0.92; on public read speech it is 0.92-0.97. Speech Commands four-way accuracy is 0.93; binary lexical ECE is at most 0.08. Personal-information detection, which passes the greedy transcript to a private, curated PII pipeline, reaches item-level F1 0.904 on synthetic Brazilian Portuguese speech (1.000 on reference transcripts). Layer-12 speaker readouts detect speaker change with AUROC 0.966-0.980 on AMI and VoxConverse, above an ECAPA baseline, and count speakers (0, 1, 2, or 3+) more accurately than pyannote 3.1. The turn-end readout reaches AUROC 0.930 versus Smart Turn v3.2 at 0.982 on Smart Turn's own test set. On a subset of ICSI meetings labeled by next speaker, AUROC is 0.716 versus 0.712; the paired confidence interval includes zero.
| Comments: | 12 pages, 4 figures, 4 tables. Code: this https URL weights: this https URL |
| Subjects: | Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Sound (cs.SD) |
| Cite as: | arXiv:2611.18330 [eess.AS] |
| (or arXiv:2611.18330v1 [eess.AS] for this version) | |
| https://doi.org/10.48550/arXiv.2611.18330 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Solano Amadori Todeschini [view email]
[v1]
Fri, 13 Nov 2026 15:44:18 UTC (318 KB)
来源:arXiv cs.CL · arxiv.org