Skip to content

Joint speech-language model - respond directly to audio!

License

Notifications You must be signed in to change notification settings

thevoicecompany/gazelle-train

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Gazelle Trainer

This is an open source trainer for Gazelle: A Joint Speech Language Model which was made by @hingeloss.

We walk through our own trainer and some data collation steps here.

You can find the trainer in the training folder along with a notebook walking you through the process.

  • This is a preliminary release for our implementation of the training codebase for Gazelle - does not include data-preprocessing
  • We are super open to collaborators and if you think you could use more code/data from us or would like to contribute to the repo - please reach out at [email protected]
  • We are training our model and it will soon be open-source!

Data formatting

  • The Data Collator takes in a HuggingFace Dataset which has to have columns "input"(text input),"audio" (list of numbers) and "text" (labels)

About

Joint speech-language model - respond directly to audio!

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 96.4%
  • Shell 3.6%