Skip to content

Latest commit

 

History

History
74 lines (46 loc) · 1.74 KB

README.md

File metadata and controls

74 lines (46 loc) · 1.74 KB
Raw

Tacotron2

Preparing

Install Dependencies

  1. Install Pytorch>=2.4
  2. Install requirements
  3. export PYTHONPATH
pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu124
pip install -r requirements.txt
export PYTHONPATH=.

Train

  1. data preprocessing
  2. run train script
python scripts/preprocessing.py -cn ljspeech
python scripts/train.py -cn ljspeech

Environment

  • OS: Windows 11, Ubuntu 22.04 LTS
  • Python: 3.11
  • Pytorch: 2.4.1
  • GPU: A40 (48gb)

Result

case 0

Text: than in the same operations with ugly ones.

case0

case 1

Text: A further development of the Roman letter took place at Venice.

case1

case 2

Text: The day room was fitted with benches and settles after the manner of the tap in a public-house.

case2

case 3

Text: This was notably the case with the early works printed at Ulm, and in a somewhat lesser degree at Augsburg.

case3

case 4

Text: result in some degree of interference with the personal liberty of those involved.

case4

case 5

Text: We have, therefore,

case5

References

  • Tacotron2 paper
  • NVIDIA Tacotron2 Implementation
  • Tacotron2 tutorial notebook