MrEdwards007

I know this plot is here for a reason but I dont know how to interpret it. My guess is that one plot is the ideal graph and the other...

I'm looking forward to looking at this tomorrow. I went for broke, and did the 'nancy' pipleline run on all 200k steps. The regression loss continued to go down, though...

Good day. I recreated my dataset with my newly created program. I found the previous version had anomalies when chunking and transcribing the data. I ran the "Nancy" pipeline for...

Reporting, so that you know that the following occurs on every long training run. In relation to the previous report, my machine is neither sleeping nor being logged out. Hopefully,...

Understood. I just need to be mindful of the cache in the future. In terms of the splitter, it can be found here: https://github.com/MrEdwards007/LJ-Speech-Dataset-Creator/tree/main I've begun using the scorer as...

>>The robotic part should be easier to fix than it not quite sounding like you. You can explore the learning rate, 1e-4 is a good default I believe there are...

> pip install torch==2.8.0 torchvision==0.23.0 torchaudio==2.8.0 --index-url https://download.pytorch.org/whl/cu129 should fix it. That worked for me.

I also needed a default or rather minimum zoom value for each waveform. I handled this by calculating the minimum pixels per second from the duration of the audio and...

I just ran into the exact same problem, where it is looking for speaker1.pkl. I backtracked through the history and saw the following. Unfortunately, restoring the "speaker1.pkl" file did not...