DuplexJev-B-Para-MOSS-Transcribe-SmolLM3-3B

A small DuplexJev connector: the MOSS-Transcribe-Diarize audio encoder (Whisper architecture, 24 layers, 307 M) (frozen) feeds a frozen SmolLM3-3B, and every typed question is read as a single-token distribution over its options, with zero decode steps and many questions per forward pass. Trained with the same recipe as the Qwen3-32B connectors released with the paper.

Quick start

pip install "duplexjev[speech]>=0.2.2"

from duplexjev import Decider, Question

d = Decider.from_pretrained("adventists-ai/DuplexJev-B-Para-MOSS-Transcribe-SmolLM3-3B")

d.decide("utterance_en.wav", [
    Question("gender", "What is the perceived gender of the speaker?", ["female", "male"], lang="en"),
    Question("emotion", "What is the speaker's emotional state?", ["neutral", "happy", "angry", "sad"], lang="en"),
], lang="en")
d.decide("utterance_zh.wav", [                            # Chinese speech: Chinese question and options
    Question("gender", "说话人的性别是？", ["男性", "女性"], lang="zh"),
    Question("emotion", "说话人当时的情绪状态是？", ["中性", "高兴", "生气", "伤心"], lang="zh"),
], lang="zh")

The connector was trained with SmolLM3's no-think format (an empty <think></think> block before the answer). The tokenizer in this repository carries that chat template; SmolLM3's default template adds a long system prompt with today's date and would not match.

The encoder is the audio encoder of MOSS-Transcribe-Diarize (Apache-2.0), repackaged as a standard Whisper encoder in adventists-ai/MOSS-Transcribe-Diarize-Whisper-Encoder.

Results

Leaderboard scores (see the GitHub README): main language (mean of qa100, ZJU-ML, Easy-Turn) 40.0, paralinguistics (mean of gender, emotion) 91.2.

Paper protocol: the evaluation prompts of the paper (single option order, one H200). duplexjev defaults: the package's own prompts with lang matched to the clip. Spoken QA with a small LLM is bounded by the LLM itself (see the transcript-reading row); use these connectors for short decisions and speaker cues rather than open knowledge questions.

Latency of one decision event (10 questions, one 4.5 s clip, duplexjev 0.2.2, plain PyTorch): 5.8 s on 8 CPU threads (Xeon Platinum 8558, fp32, not quantized); about 156 ms on one H200 (bf16, measured while the GPU was shared with a training job).

Training

Frozen encoder and LLM; only the connector is trained: R1 and R2 transcript distillation on two disjoint 0.5 M-utterance packs of the Ultravox v0.6 mixture (32k steps each, global batch 16), then 4k steps of MIX-KD on content + gender (AISHELL-1, LibriSpeech) + emotion (ESD, CREMA-D). Code and scripts: https://github.com/adventists-ai/duplexjev.

Limitations

• Evaluated on read or acted speech and small test sets (100–800 items); not on streaming input.
• Small LLMs answer knowledge questions poorly even from the transcript.
• Emotion labels come from acted corpora; do not use the outputs to make decisions about individuals.
• The connector only works with the encoder and LLM listed above.

License

CC BY-NC 4.0, research use only for the connector weights: trained on the ESD emotional speech corpus, which is licensed for research only, and on CREMA-D. Do not use it for commercial purposes. The frozen models keep their own licences: SmolLM3-3B (Apache-2.0) and the encoder (Apache-2.0).

Citation

@inproceedings{jin2027duplexjev,
  title     = {Batched Speech Decisions Without Decoding: Single-Token Supervision Lets a Frozen {LLM} Hear Beyond the Transcript},
  author    = {Jin, Jie and Ma, Ziyin and Yin, Min and Chen, Jinyu and Song, Haigang and Pang, Zhikun and Zhang, Xiaowen},
  booktitle = {Submitted to IEEE ICASSP},
  year      = {2027}
}

Built by Adventists.ai. Claude (Anthropic) assisted with code.