# machine-learning
- c
Carson
04/27/2022, 11:18 PMok - c
Carson
04/27/2022, 11:18 PMim rewatching an episode of the show so i think ill be able to get a few lines (ill make sure they have the same tone) - u
(Dawn) Will Draw Fictional Women
04/27/2022, 11:25 PMcoordinate with walt since he'll have the most up to date version of the dataset - u
user
04/28/2022, 4:24 AMi have this error " File "", line 8 warm_start, n_gpus, rank, group_name, hparams, log_directory2): ^ SyntaxError: unexpected EOF while parsing" with this code "!pwd print('FP16 Run:', hparams.fp16_run) print('Dynamic Loss Scaling:', hparams.dynamic_loss_scaling) print('Distributed Run:', hparams.distributed_run) print('cuDNN Enabled:', hparams.cudnn_enabled) print('cuDNN Benchmark:', hparams.cudnn_benchmark) def train(output_directory, log_directory, checkpoint_path, warm_start, n_gpus, rank, group_name, hparams, log_directory2):" im using Tacotron Training Notebook.ipynb and following step by step but i ran into problems - u
user
04/28/2022, 4:24 AMhow do i fix it? - g
Gosmokeless28
04/28/2022, 5:27 AMUse the Uberduck Tacotron 2 model trainer notebook: https://colab.research.google.com/drive/1WTilMdm9Vf7KE79gzkeeTBigAN6iv3Bg?usp=sharing - w
Waltman13 (semi-active)
04/28/2022, 2:48 PM@(Dawn) Will Draw Fictional Women STAR'S FINISHED!! I went through the entirety of this dataset and condensed everything down from 321 clips to just 93. Just so everything's more quality over quantity and that there isn't as much poor isolation present. had to send it in 2 parts cuz me have no nitro :(( transcript is in part 2. - u
(Dawn) Will Draw Fictional Women
04/28/2022, 2:50 PMet voila - u
user
04/28/2022, 5:43 PMim trying it now, but it says i dont have 36.npy when it is literlly there - u
user
04/28/2022, 5:44 PMalso y is it using npy? is that normal? - m
matty1406
04/28/2022, 5:45 PMtranscript problem i believe - m
matty1406
04/28/2022, 5:45 PMactually - m
matty1406
04/28/2022, 5:46 PMyou put a space in the 36 file - u
user
04/28/2022, 5:52 PMoh i didnt see that thnkas - p
PUMPKINEATER
04/28/2022, 6:19 PMTortoiseTTS is amazing - j
Justin
04/28/2022, 6:22 PMThat's actually pretty decent - z
Zachary(retired f/ voice clonin)
04/28/2022, 9:19 PMTortoiseTTS 🐢 - z
Zachary(retired f/ voice clonin)
04/28/2022, 9:24 PMooh I'll try that out - p
PUMPKINEATER
04/28/2022, 9:33 PMMore tests - u
(Dawn) Will Draw Fictional Women
04/28/2022, 9:34 PMlegit sounds like i ran a simon says call to my google home - p
PUMPKINEATER
04/28/2022, 9:40 PMI'm keeping that dataset - p
PUMPKINEATER
04/28/2022, 9:48 PMlol - c
Carson
04/28/2022, 9:57 PMi tried tortoisetts with morgan freeman - r
Reclezon
04/29/2022, 1:33 AMEclipsa model seems actually better now. Maybe a few more epochs or the Hifi tuning, and as long as you don't use
as input it'll be good.Rice is often served in round bowls. - c
Carson
04/29/2022, 1:57 AMwhats so wrong with "Rice is often served in round bowls." - c
Carson
04/29/2022, 2:00 AMin memoriam 228 wavs - c
Carson
04/29/2022, 2:01 AMim impressed though nice job waltman13 - p
PUMPKINEATER
04/29/2022, 3:08 AMMore TortoiseTTS with teenage Gilbert Gottfried