Metraipolis › archive › dc_f159eb81abfb
TEMPO: Temporally-grounded Multi-task Post-training for Large Audio-Language Models — arXiv:2608.29999
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.29999
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T07:52:21.073Z
- SHA-256 of the archived text
97e4a6e43e930231a00745476735f9b728716a286a13f220e58bdea39f68561b- Stable URL of this page
https://metraipolis.com/documento/dc_f159eb81abfb
How to cite
TEMPO: Temporally-grounded Multi-task Post-training for Large Audio-Language Models — arXiv:2608.29999 — arXiv (en), https://arxiv.org/abs/2608.29999 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f159eb81abfb (Metraipolis archive dc_f159eb81abfb, sha256 97e4a6e43e930231a00745476735f9b728716a286a13f220e58bdea39f68561b)
BibTeX
@misc{dcf159eb81abfb,
title = {TEMPO: Temporally-grounded Multi-task Post-training for Large Audio-Language Models — arXiv:2608.29999},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.29999},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f159eb81abfb; sha256 97e4a6e43e930231a00745476735f9b728716a286a13f220e58bdea39f68561b},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.29999. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Large audio-language models (LALMs) describe audio at the clip level but cannot assign timestamps to the events, speakers, or sounds they identify. Despite being essential for downstream tasks like speech recognition and dense audio captioning, timestamping remains a key limitation of most LALMs. We present TEMPO (Temporally-grounded Multi-task Post-training), the first unified model to handle audio, speech, and music timestamping tasks. Our core contribution is a supervised fine-tuning (SFT) stage built on three innovations: atomic timestamp tokens, a time-aware projector that injects sinusoidal wall-clock encodings into audio frame embeddings, and a distance-aware Gaussian loss. Our training is based on a synthetic-to-real curriculum. We further introduce, to our knowledge, the first application of reinforcement learning to unified audio timestamping, using GRPO with verifiable temporal rewards that directly optimize the evaluation objectives. Rather than serving as the primary source of performance gains, GRPO acts as a refinement stage on top of the SFT checkpoint, providing modest additional improvements. To support this work, we build a training dataset containing 119K samples and an evaluation benchmark containing 10K samples, drawn from established corpora across five tasks. On this benchmark, TEMPO outperforms Audio Flamingo Next and Qwen3-Omni, two state-of-the-art LALMs explicitly trained on timestamped data. Experiments confirm that SFT delivers most of these gains, with GRPO providing consistent but moderate refinements. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.AI). Categorie arXiv: Sound (cs.SD) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2608.29999 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-17T07:28:26.815Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).