Enregistré dans:
Détails bibliographiques
Auteurs principaux: Timmel, Vincenzo, Vogel, Manfred, Perruchoud, Daniel, Kakooee, Reza
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2506.07726
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915855444475904
author Timmel, Vincenzo
Vogel, Manfred
Perruchoud, Daniel
Kakooee, Reza
author_facet Timmel, Vincenzo
Vogel, Manfred
Perruchoud, Daniel
Kakooee, Reza
contents This paper presents a new long-form release of the Swiss Parliaments Corpus, converting entire multi-hour Swiss German debate sessions (each aligned with the official session protocols) into high-quality speech-text pairs. Our pipeline starts by transcribing all session audio into Standard German using Whisper Large-v3 under high-compute settings. We then apply a two-step GPT-4o correction process: first, GPT-4o ingests the raw Whisper output alongside the official protocols to refine misrecognitions, mainly named entities. Second, a separate GPT-4o pass evaluates each refined segment for semantic completeness. We filter out any segments whose Predicted BLEU score (derived from Whisper's average token log-probability) and GPT-4o evaluation score fall below a certain threshold. The final corpus contains 801 hours of audio, of which 555 hours pass our quality control. Compared to the original sentence-level SPC release, our long-form dataset achieves a 6-point BLEU improvement, demonstrating the power of combining robust ASR, LLM-based correction, and data-driven filtering for low-resource, domain-specific speech corpora.
format Preprint
id arxiv_https___arxiv_org_abs_2506_07726
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Swiss Parliaments Corpus Re-Imagined (SPC_R): Enhanced Transcription with RAG-based Correction and Predicted BLEU
Timmel, Vincenzo
Vogel, Manfred
Perruchoud, Daniel
Kakooee, Reza
Computation and Language
This paper presents a new long-form release of the Swiss Parliaments Corpus, converting entire multi-hour Swiss German debate sessions (each aligned with the official session protocols) into high-quality speech-text pairs. Our pipeline starts by transcribing all session audio into Standard German using Whisper Large-v3 under high-compute settings. We then apply a two-step GPT-4o correction process: first, GPT-4o ingests the raw Whisper output alongside the official protocols to refine misrecognitions, mainly named entities. Second, a separate GPT-4o pass evaluates each refined segment for semantic completeness. We filter out any segments whose Predicted BLEU score (derived from Whisper's average token log-probability) and GPT-4o evaluation score fall below a certain threshold. The final corpus contains 801 hours of audio, of which 555 hours pass our quality control. Compared to the original sentence-level SPC release, our long-form dataset achieves a 6-point BLEU improvement, demonstrating the power of combining robust ASR, LLM-based correction, and data-driven filtering for low-resource, domain-specific speech corpora.
title Swiss Parliaments Corpus Re-Imagined (SPC_R): Enhanced Transcription with RAG-based Correction and Predicted BLEU
topic Computation and Language
url https://arxiv.org/abs/2506.07726