Train TTS machine learning model
$30-250 AUD
Pago na entrega
This project is to replicate [login to view URL] Starting with a pretrained text-to-speech model trained on the LJ Speech dataset, the author was able to fine-tune the model to output the voice of a new speaker with very little additional training time (10 minutes) and very little audio training data for the new speaker (1 minute). I've tried replicating this but am getting very poor results. I have now tried several different learning rates, batch sizes, different audio data, training for different times and testing the model on different steps but in all cases the resulting audio is abysmal. I will provide SSH access to a linux server with a GPU for you to run this with, containing the pre-trained model as well as audio data I'm trying to fine-tune the model on (or you can use your own audio data - the author used voices from Modern Family characters taken from youtube). For the project to be considered completed the model must generate a voice of similar quality to the Modern Family examples given by the author.
ID do Projeto: #18256668
Sobre o projeto
5 freelancers estão ofertando em média $300 nesse trabalho
Hello, Greetings of the day.!! Your project attracted my attention at first glance, because I've really rich experience in Machine Learning with R (R studion) and Python Programming. I am having 7+ years of experi Mais
Hi, there! I am interested in your project. I have extensive experience in digital signal processing, image processing and computer vision such as speech recognition, object detection, image recognition, pattern matc Mais
Way don’t you use his model and train your with transfer learning? If you need help i more than pleased to help. As you can see i’m an expert in Deep learning technologies and Python. Regards Žiga
we have a team which has 3+ years experience in machine learning and deep learning. Your project is good. we can make it. let's discuss
