Streamalign-SLM (R16)

StreamAlign speech language model (Llama-3.2-1B backbone, R16 hier durfirst-durreg, text-KL recipe). Scores speech via the paired R16 RVQ teacher Streamalign-R16.

Speech tokenizer / teacher: Streamalign-R16
Reported metrics: SALMon 69.1 (4-task, no pitch), StoryCloze 72.1 (sSC 59.0 / tSC 85.1)

Citation

Accepted to Findings of EMNLP 2026.

@inproceedings{kim2026streamalign,
  title     = {{StreamAlign: Streaming Text-Aligned Speech Tokenization}},
  author    = {Kim, Kang-wook and Park, Jinyoung and Kim, Jinsoo and
               Lee, Sehun and Woo, Tony and Kim, Gunhee},
  booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2026},
  year      = {2026}
}