Papers for

speech therapy software developers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Speech models struggle to detect phonetic errors accurately

PEDBench: Can Speech Models Detect Phonetic Errors?

Abstract: Phonetic error detection (PED) aims to identify phoneme-level substitutions, deletions, and insertions and is essential for educational and clinical speech assessment and intervention. Despite recent advances in speech models for phoneme recognition, their ability to reliably detect fine-grained phonetic errors remains unclear because of a lack of high-quality evaluation data. To address this gap, we introduce PEDBench, the first PED benchmark spanning a continuum from natural to controlled speech. It comprises real-world recordings of spoken errors made by children (N=334), scripted error productions by human actors (N=1,188), and synthetic error productions generated using phone-conditioned text-to-speech (N=1,083). All phonetic annotations were transcribed and verified by two phonetically trained annotators. Our evaluation of 13 state-of-the-art speech models shows that they fail to reliably detect phonetic errors, even under controlled conditions with acted and synthetic speech, where acoustic variability is reduced. The models also exhibit low precision, frequently misclassifying correctly pronounced phonemes as errors. Our speech-language models, fine-tuned on synthetic data, outperform commercial and open-source baselines on word-level PED, although a sim-to-real domain gap remains. We plan to publicly release our code and recipes, along with our acted and synthetic datasets, to support the development of more accurate PED systems.

Wed 14 OctComputation and LanguageSound
The gist
Identifying small pronunciation mistakes in speech is important for helping people learn and improve their speaking. The authors created a new benchmark called PEDBench to test how well current speech models find these small errors. They collected real errors from children, acted mistakes from adults, and synthetic speech errors to make a detailed test set. Their results show that many top speech models often miss errors or falsely flag correct speech as wrong, even in easier testing conditions. They also developed models that work better using synthetic data but still don't fully solve the problem.
Open → 2610.20830v1

Audio language models struggle with child stuttering speech understanding

Reasoning Beyond Transcription: Audio Language Models on Child Stuttering Speech

Abstract: Child speech differs from adult speech in acoustics, prosody, and linguistic structures. Speech disfluencies (such as repetitions) further challenge automatic understanding. While Audio Language Models (ALMs) show strong semantic reasoning from speech audio, their ability to reason about disfluent child speech in mixed-speaker settings remains unexplored. We investigate this through two tasks: child-focused semantic summarization and speech entailment. Experiments use recordings of children who stutter in mixed speaker interviews without explicit speaker separation. Models are instruction-guided to focus on the child, preserve clinically relevant disfluencies, and avoid adult-speech leakage. Evaluation combines LLM-based judges and reference-based metrics, anchored by transcript-oracle baselines to isolate errors. Results show that while ALMs extract high-level meaning from stuttered speech, reasoning degrades significantly with increased

Mon 7 SeptComputation and Language
The gist
Understanding speech from children who stutter is hard because their speech sounds, rhythms, and word patterns are different and often interrupted with repetitions. The authors studied if special audio language models can still make sense of what these children say, even when adults are also speaking. They tested if the models could summarize or judge sentences focused on the children, keeping important stuttering details. The results showed the models do understand the main ideas but have trouble when the stuttering or mixed voices increase.
Open → 2609.07968v1