Sign in to view source links and access this dataset
Description
A cleaned Persian speech dataset built from the FLEURS Persian subset. The audio has been processed with the Representation Chizzler pipeline, which includes segmentation, denoising, loudness normalization, and resampling. The dataset was authored by Reza2kn and last updated on January 11, 2026.
Use Cases
Train or fine-tune Persian automatic speech recognition (ASR) models based on the cleaned speech audio.
Benchmark audio denoising and segmentation algorithms based on the processed audio samples.
Develop speech representation models for Persian based on the normalized and segmented audio.
Conduct research on multilingual speech models using a cleaned Persian speech corpus.
Strengths
Audio has been processed through a specific pipeline (Representation Chizzler) for cleaning.
Processing steps include speech segmentation with Silero VAD, denoising with MP-SENet, loudness normalization, and resampling to 16 kHz.
Dataset is derived from the established FLEURS Persian subset, providing a known base source.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count, file formats, and dataset size are unknown, which may limit suitability assessment.
License information is unknown, which could restrict commercial or research use.
Provenance
Source
Original data from the FLEURS Persian subset on Hugging Face.
Collection Method
Processed with the Representation Chizzler pipeline (Silero VAD, MP-SENet denoising, loudness normalization, 16 kHz resampling).
Time Range
null
Freshness
Last updated 2026-01-11 08:17:37; freshness should be verified.
Geography
null
License restrictions are unknown; users should verify before use.