After more than five months of development, version 1.7.0 of the high-performance speech recognition system has been released. whisper.cpp, implementing the model Whisper by OpenAI in C and C++, distributed under the MIT license.
The project utilizes a custom machine learning tensor library GGML, written in C.
Key Changes:
- crashes with a large number of beams have been fixed;
- overall VRAM usage has been reduced;
- encoder performance has been optimized.
Source: linux.org.ru
