Coder Social home page Coder Social logo

jets's Introduction

Introduction

  1. FastSpeech2, HiFi-GAN 오픈 소스를 활용하여 JETS(End-To-End)를 간단 구현하고 한국어 데이터셋(KSS)을 사용해 빠르게 학습합니다.
  2. 기존 오픈소스는 MFA기반 preprocessing을 진행한 상태에서 학습을 진행하지만 본 레포지토리에서는 alignment learning 기반 학습을 진행하고 preprocessing으로 인해 발생할 수 있는 디스크 용량 문제를 방지하기 위해 data_utils.py로부터 학습 데이터가 feeding됩니다.
  3. conda 환경으로 진행해도 무방하지만 본 레포지토리에서는 docker 환경만 제공합니다. 기본적으로 ubuntu에 docker, nvidia-docker가 설치되었다고 가정합니다.
  4. GPU, CUDA 종류에 따라 Dockerfile 상단 torch image 수정이 필요할 수도 있습니다.
  5. preprocessing 단계에서는 학습에 필요한 transcript와 stats 정도만 추출하는 과정만 포함되어 있습니다.
  6. 그 외의 다른 preprocessing 과정은 필요하지 않습니다.
  7. End-To-End & Adversarial training 기반이기 때문에 우수한 품질의 오디오를 생성하기 위해선 많은 학습을 필요로 합니다.

Dataset

  1. download dataset - https://www.kaggle.com/datasets/bryanpark/korean-single-speaker-speech-dataset
  2. unzip /path/to/the/kss.zip -d /path/to/the/kss
  3. mkdir /path/to/the/JETS/data/dataset
  4. mv /path/to/the/kss.zip /path/to/the/JETS/data/dataset

Docker build

  1. cd /path/to/the/JETS
  2. docker build --tag JETS:latest .

Training

  1. nvidia-docker run -it --name 'JETS' -v /path/to/JETS:/home/work/JETS --ipc=host --privileged JETS:latest
  2. cd /home/work/JETS
  3. ln -s /home/work/JETS/data/dataset/kss
  4. python preprocess.py ./config/kss/preprocess.yaml
  5. python train.py -p ./config/kss/preprocess.yaml -s ./config/kss/model.yaml -g ./config/kss/config_v1.json -t ./config/kss/train.yaml
  6. arguments
  • -p : preprocess config path
  • -s : synthesizer config path
  • -g : generator config path
  • -t : train config path
  1. (OPTIONAL) tensorboard --logdir=outdir/logdir

Tensorboard losses

JETS-tensorboard-losses

Tensorboard Stats

JETS-tensorboard-images

Reference

  1. JETS: Jointly Training FastSpeech2 and HiFi-GAN for End to End Text to Speech
  2. FastSpeech2 github
  3. Comprehensive-Transformer-TTS
  4. HiFi-GAN
  5. VITS

jets's People

Contributors

ailab-choihk avatar choihkk avatar

Stargazers

MO_DEV avatar Ven_ avatar  avatar  avatar shenquanbo avatar  avatar Nickolay V. Shmyrev avatar Rishikesh (ऋषिकेश) avatar Yuan-Man avatar  avatar  avatar Qian Liu avatar zyser avatar

Watchers

Nickolay V. Shmyrev avatar zyser avatar  avatar

jets's Issues

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.