Go to file
2023-07-26 13:57:32 +02:00
.vscode ENH Forward pass text-encoder (#3) 2023-07-14 10:50:44 +02:00
.gitignore UPD gitignore 2023-07-09 00:02:19 +02:00
bark.cpp FEAT Add forward pass of coarse model (#5) 2023-07-16 23:07:29 +02:00
bark.h FEAT Add forward pass of coarse model (#5) 2023-07-16 23:07:29 +02:00
convert_pt_to_ggml.py FEAT Add WordPiece BERT Tokenization (#1) 2023-07-09 19:44:00 +02:00
encodec.cpp working loading weights 2023-07-09 10:32:36 +02:00
encodec.h working loading weights 2023-07-09 10:32:36 +02:00
ggml.c working convert_pt_to_ggml script 2023-07-08 15:54:41 +02:00
ggml.h working convert_pt_to_ggml script 2023-07-08 15:54:41 +02:00
Makefile Make recipes 2023-07-09 00:01:25 +02:00
README.md DOC README file (#9) 2023-07-26 13:57:32 +02:00
util.h FEAT Add WordPiece BERT Tokenization (#1) 2023-07-09 19:44:00 +02:00

bark.cpp (coming soon!)

Inference of SunoAI's bark model in pure C/C++ using ggml.

Description

The main goal of bark.cpp is to synthesize audio from a textual input with the Bark model using 4-bit quantization on a MacBook.

Bark has essentially 4 components:

  • Semantic model to encode the text input
  • Coarse model
  • Fine model
  • Encoder (quantizer + decoder) to generate the waveform from the tokens

Roadmap

  • Quantization
  • FP16
  • Swift package for iOS devices