cs.CLSep 13, 2026

Neyshekar: An Open Persian Read-Speech Corpus for Automatic Speech Recognition

Authors: Ahmad Amirivojdan, Farzad Nadiri, Abolfazl Alizadeh, Shaghayegh Yaraghi

Organizations: Shekar AI

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

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.

Figures & tables

CorpusSpeech and sourceHoursSpeakers
FARSDATRead; 10 dialect regions–300
MirasVoiceRead + spontaneous Fa–En>> 3350
ShEMOSemi-natural emotional3.487
DeepMineFa–En recordings>> 480>> 1.8k
ManaTTSSingle-speaker narration861
ParsVoiceAligned audiobook segs.2,2001,815
Table 1: Selected Persian speech corpora related to Neyshekar. Values are taken from published corpus descriptions, except for Common Voice and Neyshekar, for which values were computed from the analysed releases.
Figure 1: Approximate prompt-pool composition reported by the authors. Shares refer to prompt sources, not recording counts or audio duration; generated sources together account for 82% of the pool.
MeasureValue
Independent labels1,200
Accept-label share86.3%
Fleiss’ κ\kappa (95% CI)0.679 [0.591, 0.757]
Gwet’s AC1 (95% CI)0.900 [0.866, 0.930]
Overall raw agreement92.4%
Agreement on accept95.6%
Table 2: Inter-annotator reliability on the shared sample of 300 recordings (four independent annotators; 95% CIs).
Figure 2: Clip-duration distributions for Neyshekar v6 and validated Persian Common Voice 26.0 (proportion of all clips). The 88 Neyshekar clips and two Common Voice clips longer than 20 s (0.14% and under 0.01% of each corpus) are omitted rather than collected into the final bin.
Figure 3: Cross-corpus performance after three-epoch training on frozen approximately 32h subsets. WER and CER are shown separately. Small dots denote individual optimisation seeds; diamonds and whiskers denote the mean and sample SD across three seeds, not confidence intervals. Lower values indicate lower error rates.
Figure 4: External speech evaluation of the three-epoch, approximately 32h fine-tuned systems. Small dots denote individual seeds; diamonds and whiskers denote the three-seed mean and sample SD. PSRB is restricted to the 344-clip public sample. Lower WER/CER is better.
Figure 5: Whisper small scaling under three epochs and a fixed 1,425-update budget. Actual post-filtering durations from frozen manifests are plotted on a logarithmic horizontal axis. Only optimisation seed 42 is represented; uncertainty bands are therefore not provided. Equal updates do not imply equal FLOPs or audio exposure.
Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Figure A1: The 64h mixture control at 2,382 updates: mixed minus Neyshekar-only WER/CER in percentage points. Negative differences favour mixing. Each point and interval represents one optimisation seed and its paired percentile 95% bootstrap interval. Utterances are resampled within each test. Intervals are conditional on trained systems and are not multiplicity-adjusted.
More in this topic

Automatic Speech Recognition

Automatic speech recognition converts spoken language into text across speakers, languages, domains, and recording conditions.

See all 480 papers

Explore similar work

CardsList
Oct 12, 2025

ParsVoice: A Large-Scale Multi-Speaker Persian Speech Corpus for Text-to-Speech Synthesis

Persian remains substantially underrepresented in open speech-text resources, limiting progress in multi-speaker text-to-speech (TTS), speech-language modelling, and low-resource speech processing. We introduce ParsVoice, the largest publicly available Persian speech-text corpus tailored for training multi-speaker TTS systems, along with a scalable pipeline to construct high-quality speech-text data from long-form audiobook recordings. The pipeline combines a fine-tuned ParsBERT sentence-completion classifier, ASR-based boundary optimization, punctuation restoration, speaker identification, and a multi-dimensional quality assessment that covers both audio and Persian-specific text properties. The resulting release contains a 2,200-hour TTS-ready subset with 1.36 million aligned segments from 1,815 automatically inferred speaker IDs, making it more than 25 times larger than the largest previously available open Persian TTS dataset. To validate the corpus, we fine-tune XTTSv2, a zero-shot multilingual TTS model that operates directly on raw Persian text without phoneme representations. The resulting model achieves a naturalness MOS of 3.6/5 and a speaker-similarity MOS of 4.0/5. ParsVoice, its metadata, and the corpus-construction pipeline are publicly available on Hugging Face: https://huggingface.co/datasets/MohammadJRanjbar/ParsVoice and GitHub: https://github.com/MohammadJRanjbar/ParsVoice, supporting reproducible research on Persian speech synthesis and low-resource speech-language technologies.

Mohammad Javad Ranjbar Kalahroodi, Heshaam Faili, Azadeh Shakery

School of Electrical and Computer Engineering, University of Tehran, Iran · Institute for Research in Fundamental Sciences (IPM), Tehran, Iran

TTS SynthesisMulti-Speaker TTSSpeech ProcessingSpeech Corpus Construction
AskPDF
Mar 9, 2026

Ramsa: A Large Sociolinguistically Rich Emirati Arabic Speech Corpus for ASR and TTS

Ramsa is a developing 41-hour speech corpus of Emirati Arabic designed to support sociolinguistic research and low-resource language technologies. It contains recordings from structured interviews with native speakers and episodes from national television shows. The corpus features 157 speakers (59 female, 98 male), spans subdialects such as Urban, Bedouin, and Mountain/Shihhi, and covers topics such as cultural heritage, agriculture and sustainability, daily life, professional trajectories, and architecture. It consists of 91 monologic and 79 dialogic recordings, varying in length and recording conditions. A 10% subset was used to evaluate commercial and open-source models for automatic speech recognition (ASR) and text-to-speech (TTS) in a zero-shot setting to establish initial baselines. Whisper-large-v3-turbo achieved the best ASR performance, with average word and character error rates of 0.268 and 0.144, respectively. MMS-TTS-Ara reported the best mean word and character rates of 0.285 and 0.081, respectively, for TTS. These baselines are competitive but leave substantial room for improvement. The paper highlights the challenges encountered and provides directions for future work.

Rania Al-Sabbagh

Department of Foreign Languages, University of Sharjah, UAE

Speech Corpus ConstructionArabic NLPArabic DialectsLow-Resource Speech RecognitionLow-Resource TTS Synthesis
AskPDF
Mar 27, 2026

Pashto Common Voice: Building the First Open Speech Corpus for a 60-Million-Speaker Low-Resource Language

We present the Pashto Common Voice corpus -- the first large-scale, openly licensed speech resource for Pashto, a language with over 60 million native speakers largely absent from open speech technology. Through a community effort spanning 2022-2025, the corpus grew from 1.5 hours and 5 contributors to 147 total hours and 1,483 unique speakers across ten Mozilla Common Voice releases (CV14-CV23). Speaker participation increased approximately 108-fold between CV17 and CV18, coinciding with a VOA Pashto broadcast campaign. We describe the full methodology: interface localisation, Wikipedia-based sentence extraction with automated filtering, phonemically targeted contributions for the four most frequently dropped Pashto characters, and multi-channel community outreach. MCV23 contains 107,781 clips (60,337 validated; 82.33 validated hours) across 13 content domains. Fine-tuning Whisper Base on the MCV20 yields 13.4% WER on the MCV20 test split, against the published Whisper Base zero-shot WER of 99.0% on Pashto.

Hanif Rahman, Shafeeq ur Rehman

Pashto DAO

Speech Corpus ConstructionLow-Resource Language ProcessingSpeech ProcessingLow-Resource Speech Recognition
AskPDF