FOCAL: A Novel Benchmarking Technique for Multi-modal Agents

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Purwar, Anupam, Choudhary, Aditya
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915826905382912
author Purwar, Anupam
Choudhary, Aditya
author_facet Purwar, Anupam
Choudhary, Aditya
contents With the recent advancements in reasoning capabilities, tool calling using MCP servers and Audio Language Models (ALMs), development and integration of multi-modal agents (with voice and text support) has come to the industry forefront. Cascading pipelines for voice agents still play a central role in the industry owing to their superior reasoning capabilities facilitated by LLMs. Although, cascading pipelines often present error propagation through the pipeline. We propose a framework, FOCAL to benchmark end-to-end reasoning, component-wise error propagation and error analysis for automated as well as human-assisted testing of multi-modal agents (voice to voice + text input). We also share two novel metrics viz. Reasoning and Semantic scores to evaluate efficacy of the agent in having meaningful conversations in voice mode.
format Preprint
id arxiv_https___arxiv_org_abs_2601_07367
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle FOCAL: A Novel Benchmarking Technique for Multi-modal Agents
Purwar, Anupam
Choudhary, Aditya
Sound
With the recent advancements in reasoning capabilities, tool calling using MCP servers and Audio Language Models (ALMs), development and integration of multi-modal agents (with voice and text support) has come to the industry forefront. Cascading pipelines for voice agents still play a central role in the industry owing to their superior reasoning capabilities facilitated by LLMs. Although, cascading pipelines often present error propagation through the pipeline. We propose a framework, FOCAL to benchmark end-to-end reasoning, component-wise error propagation and error analysis for automated as well as human-assisted testing of multi-modal agents (voice to voice + text input). We also share two novel metrics viz. Reasoning and Semantic scores to evaluate efficacy of the agent in having meaningful conversations in voice mode.
title FOCAL: A Novel Benchmarking Technique for Multi-modal Agents
topic Sound
url https://arxiv.org/abs/2601.07367