search

hirofumi0810 / tensorflow_end2end_speech_recognition

End-to-End speech recognition implementation base on TensorFlow (CTC, Attention, and MTL training)
MIT License
312 stars 120 forks source link

Is this project still under development? #5

Closed hlthu closed 7 years ago

hlthu commented 7 years ago

Hello, I am a student of first year of Master in China, and I am interested in speech recognition and your project. Is this project still under development?

I have used Kaldi in the past, and is still using it, but I think end-to-end will have a brighter future! So I hope that I can cooperate with you.

Thx

hirofumi0810 commented 7 years ago

Hello,

I'm glad to hear that!

My goal is to replace Kaldi toolkit with python-based deep learning toolkit such as Tensorflow, chainer, or Pytorch. So i think your experience in Kaldi will help me.

Now i'm working on the research internship, so i don't have enough time to maintain this repo by myself. Please help me!

Future works are as follows:

  1. Introduce RNNLM (char and word) or N-gramLM
  2. WFST decoding with CTC or Attention-based model
  3. Sequence training such as sMBR
  4. Add other corpora
  5. (Online implementation)
  6. (Merge with asr_preprocessing??)

thank you!

Hiforumi

2017-09-13 2:01 GMT+09:00 Lu Huang notifications@github.com:

Hello, I am a student of first year of Master in China, and I am interested in speech recognition and your project. Is this project still under development?

I have used Kaldi in the past, and is still using it, but I think end-to-end will have a brighter future! So I hope that I can cooperate with you.

Thx

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/hirofumi0810/tensorflow_end2end_speech_recognition/issues/5, or mute the thread https://github.com/notifications/unsubscribe-auth/ARGAg76f6KvKMHuSW79IZcrfOrPXYgmpks5shrjbgaJpZM4PU9kp .

hlthu commented 7 years ago

Ok, I will first get familiar to your code~ Thx