---
type: "post"
id: "new-paper-proposes-abstract-compression-for-conversational-context-b183c5"
slug: "new-paper-proposes-abstract-compression-for-conversational-context-b183c5"
title: "New paper proposes Abstract Compression for conversational context in LLM-based ASR."
description: "A paper by Shashi Kumar and nine coauthors proposes “Abstract Compression” for conversational context in LLM-based automatic speech recognition."
retrieval_nugget: "A paper by Shashi Kumar and nine coauthors proposes “Abstract Compression” for conversational context in LLM-based automatic speech recognition."
published_at: "2026-10-23"
updated_at: "2026-10-23"
record_date: "2026-10-23"
date_kind: "published_at"
topics: ["context-engineering"]
entities: ["Abstract Compression","LLM-based ASR","IEEE SLT 2026","arXiv","Shashi Kumar"]
related_patterns: []
novelty_label: "new_fact"
what_changed: "A new ASR paper introduces fixed-token latent compression of prior-turn audio, retaining transcripts and recovering some raw-context gains; the priors cover different LLM representation and video-token compression research."
prior_refs: ["memory:752be034-060a-800c-a9d3-fc2bdc757b0a","memory:d00ca9a1-a805-86e9-af9e-016e142238a5"]
source_url: "https://arxiv.org/abs/2603.26246"
source_urls: ["https://arxiv.org/abs/2603.26246"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:new-paper-proposes-abstract-compression-for-conversational-context-b183c5"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/new-paper-proposes-abstract-compression-for-conversational-context-b183c5/","markdown":"https://newruntime.com/posts/new-paper-proposes-abstract-compression-for-conversational-context-b183c5.md","json":"https://newruntime.com/posts/new-paper-proposes-abstract-compression-for-conversational-context-b183c5.json"}
---

# New paper proposes Abstract Compression for conversational context in LLM-based ASR.

## Retrieval answer

A paper by Shashi Kumar and nine coauthors proposes “Abstract Compression” for conversational context in LLM-based automatic speech recognition.

A paper by Shashi Kumar and nine coauthors proposes “Abstract Compression” for conversational context in LLM-based automatic speech recognition. The current arXiv version was revised on October 20, 2026, and the paper is listed as accepted to IEEE SLT 2026.

The authors study prior-turn context in multi-turn speech recognition. They report that, after supervised multi-turn training, conversational context mainly improves recognition of contextual entities. Raw prior-turn audio embeddings grow with conversation length, so the method replaces each prior audio sequence with a fixed number of learned latent tokens while keeping the corresponding transcripts explicitly available.

On in-domain and out-of-domain test sets, the authors say the compression-enabled model recovers part of the gain from raw-context conditioning with a smaller prior-turn audio footprint. The abstract does not provide absolute metrics, implementation cost, code or data availability, or evidence that the method is ready for production use.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Context Engineering](https://newruntime.com/topics/context-engineering.md): Explore the context-engineering topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)