callmephilip

0 Followers · 0 Following

• Joined on 2023-10-09

Repositories Projects Packages Public Activity Starred Repositories

callmephilip opened issue ecker/ai-voice-cloning#423 2023-10-21 11:19:15 +00:00

20 min audio, 500 epochs, underwhelming result

callmephilip commented on issue ecker/ai-voice-cloning#405 2023-10-20 11:06:54 +00:00

Does not run in Docker

@mrq i can confirm that the latest Dockerfile seems to work fine. I ran into a different problem training on a larger audio (~20 min) where PyTorch DataLoader was running out of memory. the…

callmephilip commented on issue ecker/ai-voice-cloning#405 2023-10-10 14:56:56 +00:00

Does not run in Docker

ok. your miniconda comment got me thinking. here's an updated version of the Docker file that seems to work for me. using Cuda 11.7

FROM nvidia/cuda:11.7.1-cudnn8-devel-ubuntu20.04

…

callmephilip commented on issue ecker/ai-voice-cloning#405 2023-10-09 17:19:20 +00:00

Does not run in Docker

Reran with the following updated Dockerfile:

FROM nvidia/cuda:11.7.1-cudnn8-devel-ubuntu20.04

ARG DEBIAN_FRONTEND=noninteractive
ARG TZ=UTC
ARG MINICONDA_VERSION=23.1.0-1
ARG…

callmephilip opened issue ecker/ai-voice-cloning#405 2023-10-09 17:01:36 +00:00

Does not run in Docker