S-DiverSe: Spanish Diverse Speech

2026-07-03

AI Summary

This study addresses the significant performance degradation of existing Spanish automatic speech recognition (ASR) systems on speech from individuals with neurodegenerative diseases, primarily due to a lack of diverse, real-world evaluation data. To bridge this gap, the authors construct and publicly release the first in-the-wild Spanish speech dataset targeting speakers with amyotrophic lateral sclerosis, Parkinson’s disease, and post-stroke conditions. The dataset comprises 3.2 hours of natural conversational speech from 22 speakers, accompanied by 444 manually transcribed utterances and rich metadata including speaker gender, disease type, and intelligibility ratings. Using this resource, the authors establish baseline ASR systems and conduct adaptation experiments, finding that heuristic text-based post-processing outperforms model fine-tuning for cross-domain recognition of dysarthric speech, thereby demonstrating the critical value of domain-specific, real-world data in enhancing ASR robustness.

Abstract

Automatic speech recognition (ASR) has advanced remarkably for standard speech, yet speech affected by neurological conditions remains a challenge. We present S-DiverSe (Spanish Diverse Speech), a corpus of 3.2 hours of in-the-wild Spanish speech from 22 speakers with amyotrophic lateral sclerosis, Parkinson's disease, and stroke. The dataset contains 444 manually transcribed audio segments with metadata on speaker sex, disease type, and intelligibility. S-DiverSe is designed to support ASR evaluation and development for neurologically affected Spanish speech. We describe the dataset, analyze its composition, and report baseline ASR results alongside initial adaptation experiments. Our findings reveal that heuristic text post-processing is more robust than fine-tuning for out-of-domain neurological Spanish speech. This underscores the need for dedicated in-the-wild Spanish benchmarks.

Problem

• automatic speech recognition
• neurological speech disorders
• Spanish speech
• in-the-wild dataset
• speech intelligibility

Innovation

• Spanish Diverse Speech
• neurological speech disorders
• in-the-wild ASR
• text post-processing
• speech corpus