I have couple of audio recordings of length 5 -7 min. I also have corresponding text to each audio file. I am not subject matter expert on speech to text but I am vaguely aware that much smaller audio files (10 sec) are used to fine-tune speech recognition model . So I want to split audio files into meaningful partitions . Correct me if I am wrong in assumption that I cannot train model with large audio files.
1 Answer
$\begingroup$
$\endgroup$
1
Some toolkits like Kaldi can handle longer inputs. Others indeed require 10 seconds or so. It depends on the tools you gonna use.
-
$\begingroup$ I am definitely going to split audio files because i have very low no of files already , i thought i could increase samples by audio segmentation.I am actually collecting this dataset to fine tune model to Asian Indian Accent.Also there is limit to amount of hours of audio that i can collect. $\endgroup$ Mar 31, 2019 at 19:59