VoxRAG: A Step Toward Transcription-Free RAG Systems in Spoken Question Answering

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Rackauckas, Zackary, Hirschberg, Julia
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866908480642744320
author Rackauckas, Zackary
Hirschberg, Julia
author_facet Rackauckas, Zackary
Hirschberg, Julia
contents We introduce VoxRAG, a modular speech-to-speech retrieval-augmented generation system that bypasses transcription to retrieve semantically relevant audio segments directly from spoken queries. VoxRAG employs silence-aware segmentation, speaker diarization, CLAP audio embeddings, and FAISS retrieval using L2-normalized cosine similarity. We construct a 50-query test set recorded as spoken input by a native English speaker. Retrieval quality was evaluated using LLM-as-a-judge annotations. For very relevant segments, cosine similarity achieved a Recall@10 of 0.34. For somewhat relevant segments, Recall@10 rose to 0.60 and nDCG@10 to 0.27, highlighting strong topical alignment. Answer quality was judged on a 0--2 scale across relevance, accuracy, completeness, and precision, with mean scores of 0.84, 0.58, 0.56, and 0.46 respectively. While precision and retrieval quality remain key limitations, VoxRAG shows that transcription-free speech-to-speech retrieval is feasible in RAG systems.
format Preprint
id arxiv_https___arxiv_org_abs_2505_17326
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle VoxRAG: A Step Toward Transcription-Free RAG Systems in Spoken Question Answering
Rackauckas, Zackary
Hirschberg, Julia
Information Retrieval
Sound
Audio and Speech Processing
We introduce VoxRAG, a modular speech-to-speech retrieval-augmented generation system that bypasses transcription to retrieve semantically relevant audio segments directly from spoken queries. VoxRAG employs silence-aware segmentation, speaker diarization, CLAP audio embeddings, and FAISS retrieval using L2-normalized cosine similarity. We construct a 50-query test set recorded as spoken input by a native English speaker. Retrieval quality was evaluated using LLM-as-a-judge annotations. For very relevant segments, cosine similarity achieved a Recall@10 of 0.34. For somewhat relevant segments, Recall@10 rose to 0.60 and nDCG@10 to 0.27, highlighting strong topical alignment. Answer quality was judged on a 0--2 scale across relevance, accuracy, completeness, and precision, with mean scores of 0.84, 0.58, 0.56, and 0.46 respectively. While precision and retrieval quality remain key limitations, VoxRAG shows that transcription-free speech-to-speech retrieval is feasible in RAG systems.
title VoxRAG: A Step Toward Transcription-Free RAG Systems in Spoken Question Answering
topic Information Retrieval
Sound
Audio and Speech Processing
url https://arxiv.org/abs/2505.17326