# machine-learning
- t
Toasty
04/30/2021, 8:23 PMWait - t
Toasty
04/30/2021, 8:23 PMI have to remake the dataset? - u
user
04/30/2021, 8:24 PMI mean it should be a case of just importing all the files into audacity, pitching them 0.5 semitones up and re-exporting - t
Toasty
04/30/2021, 8:24 PMAlright. - t
Toasty
04/30/2021, 8:24 PMI can try that. - t
Toasty
04/30/2021, 8:24 PMWorse case, if that doesn't work, I'll just have to buy a copy of his audiobook - t
Toasty
04/30/2021, 8:25 PMOr find a different source. - t
Toasty
04/30/2021, 8:25 PMJames May definitely is behind a paywall. - u
user
04/30/2021, 8:26 PMThe speed might or might not also be off depending, lemme quickly check that for you - u
user
04/30/2021, 8:40 PMOk to my knowledge the speed isnt off - u
user
04/30/2021, 8:41 PMI think you might also just want to use different sentences really, as there's three hours of data - u
(Edd) bruh moment
04/30/2021, 9:08 PMI'm laughing - t
Toasty
04/30/2021, 9:31 PMWell, I used an automated dataset maker - t
Toasty
04/30/2021, 9:31 PMIdk what the program chose. - u
user
04/30/2021, 9:31 PMTacotron2AutoTrim? - t
Toasty
04/30/2021, 9:32 PMNo - t
Toasty
04/30/2021, 9:33 PMI put the 3 hour audio file in with the sample - t
Toasty
04/30/2021, 9:34 PMAny use the sample to weed out the good idea with the bad audio - t
Toasty
04/30/2021, 9:35 PMIt did a good job. Got 2 hours of audio out of it. - u
user
04/30/2021, 9:35 PMDoes it build a dataset - t
Toasty
04/30/2021, 9:35 PMYes. - u
user
04/30/2021, 9:35 PMLike wavs - t
Toasty
04/30/2021, 9:35 PMBuilds a dataset and transcript with arpabet. - u
user
04/30/2021, 9:35 PMDoes it need cuda - u
user
04/30/2021, 9:35 PM? - t
Toasty
04/30/2021, 9:35 PMIdk. - t
Toasty
04/30/2021, 9:35 PMI used Google Colab. - t
Toasty
04/30/2021, 9:36 PMIt uses IBM's services tho also - u
user
04/30/2021, 9:36 PMWhere is the repo? - t
Toasty
04/30/2021, 9:37 PMIdk.