Ten single-speaker synthetic Najdi Arabic TTS sets unified into one dataset, distinguished by the speaker column. 91,350 clips across 10 speaker identities, 24 kHz mono. This consolidates the former Salesteq/ar-sa-tts-speaker{1-10} repos into a single dataset — filter by speaker for one voice.
Sentences are synthesised with OmniVoice (Salesteq/omnivoice-ar-sa-saudi, language="ars"… See the full description on the dataset page:
https://huggingface.co/datasets/Salesteq/ar-sa-tts-speakers-synthetic.
