Urdu TTS Mini: Speech Audio and Text for Urdu Language Models

A curated Urdu speech dataset for Text-to-Speech and Automatic Speech Recognition research. Audio segments are extracted from publicly available YouTube speech content, processed through a multi-stage quality pipeline, and annotated with Urdu transcriptions. This mini release by salisai, last updated in June 2026, is intended to validate the preprocessing pipeline and establish a quality baseline for future large-scale versions.

Use Cases

• Train Urdu text-to-speech models based on the paired audio and transcription data.
• Fine-tune Urdu automatic speech recognition systems based on the annotated speech segments.
• Validate audio preprocessing and quality assurance pipelines for speech data.
• Establish baseline performance metrics for Urdu speech technology research.

Strengths

• Data is curated and processed through a described multi-stage quality pipeline.
• Transcriptions are provided in Urdu, enabling supervised learning tasks.
• The dataset is described as a 'mini release' intended to establish a quality baseline.

Limitations

• Dataset size, row count, and file formats are unknown, limiting suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect source bias inherent to the YouTube content used for extraction.

Provenance

Source

Publicly available YouTube speech content.

Collection Method

Audio segments extracted and processed through a multi-stage quality pipeline.

Freshness

Last updated 2026-06-16 09:51:34