Datasets:
Dataset haqida
#3
by
yakhyo
- opened
Assalomu alaykum, chiqarilgan dataset uchun katta rahmat. Men quyidagi savollar bor edi dataset bo'yicha, javob bersangiz juda hursand bo'lar edim.
Bu dataset Mozilla Common Voice dataseti bilan ustma ust tushmaydimi? Sababi men hozir Mozilla Common Voice ni ishlatayotgan edim, agarda bu boshqa dataset bo'lsa, buni ustida ham training qilmoqchi edim.
Bu yerdagi ma'lumotlar barchasi filterlab chiqilganmi? Nomida yozilgan ekan, lekin agarda ishlatishdan oldingi biror tafsiyalariz bo'lsa , ulashsangiz yaxshi bo'lar edi.
Kattakon rahmat.
- filterlangan deganda nimasi nazarda tutilyapti? audiolar validatsiya qilinganmi shunda? (eshitib, to'g'ri o'qilganligi tekshirilganmi)
yoki bo'sh qatorlar olib tashlanganmi?
We filtered the dataset using number approaches:
- VAD
- Reading Speed
- Automatic STT checking - we trained the model using subset of samples from different authors that we approved and used trained model to extend the number of samples given their transcription match our trained model output to some extend
updated Readme