Neyshekar open Persian speech dataset improves language recognition
Neyshekar: An Open Persian Read-Speech Corpus for Automatic Speech Recognition
Computation and LanguageSound
Summary
Understanding spoken Persian by computers requires lots of recorded examples of people speaking. The authors created Neyshekar, a large collection of recorded Persian speech covering both formal and informal language from many speakers. They carefully checked the recordings and the text for accuracy and tested how well computers could recognize speech using this data. The new dataset helped models recognize Persian speech better than an existing open dataset of similar size.
What this means in practice
- •For speech recognition developers: Train Persian speech recognition systems with a large variety of formal and informal speech for improved accuracy, using a validated open dataset.
- •For language technology teams: Evaluate and improve Persian language models with a dataset that includes named entities and longer utterances, enabling better natural language understanding.
Authors
Ahmad Amirivojdan, Farzad Nadiri, Abolfazl Alizadeh, Shaghayegh Yaraghi
Abstract
Neyshekar is presented as an open Persian read-speech corpus designed for coverage of both formal and informal language, named entities, and longer utterances. In version 6, 62,279 validated recordings totalling 99.02 hours are provided from 190 contributors, with 34,541 distinct recorded prompts. The prompt pool was assembled from human-written material, contextualised homographs, and reviewed language-model-generated text. Text entries were normalised with the shekar library, which supports both formal and informal Persian, and every submitted recording was reviewed against a common validation rubric. About 24% of released clips are classified as informal by an automatic classifier; these register labels are not human-validated. Item-level rater labels are provided for reproducible agreement estimation, opaque per-clip contributor identifiers make the speaker-disjoint partitioning auditable and support contributor-clustered uncertainty estimates, and a text-disjoint test subset is included for evaluation beyond previously seen prompts. Per-contributor recording load and reference-free signal quality are characterised for every released clip. Corpus characteristics are compared with Persian Common Voice under shared processing. Utility is assessed through two ASR architectures, three optimisation seeds, WER and CER, and independent evaluation on the public PSRB sample. Against duration-matched Common Voice training at approximately 32 hours, in-domain WER is reduced by 9.5 points for Whisper and 11.6 points for XLS-R, and by approximately eight points for both architectures on the independent PSRB sample. Transfer and mixture benefits are not consistently observed across architectures and training budgets. The corpus is released under CC0; code and data are made available through the project repository at https://github.com/amirivojdan/neyshekar.