handy-computer / transcribe.cpp

TL;DR

A C/C++ speech-to-text (STT) inference library has been developed to run various STT model families using GGUF models on the ggml runtime. It supports fast GPU inference through Metal, Vulkan, and CUDA, along with a CPU path accelerated by BLAS.

✦ Why It Matters

Engineers can choose from 16 model families to optimize speech recognition for their specific application needs today.

Key Takeaways

Explore 16 model families for diverse transcription needs.
Utilize GPU acceleration with Metal, Vulkan, or CUDA.
Build and test models using CMake for easy integration.
Leverage quantization tools to optimize model sizes.

How It Works

Transcribe.cpp operates by leveraging the ggml runtime to execute various speech-to-text models efficiently. It supports multiple hardware backends, allowing for optimized performance on different platforms.

The library's architecture enables both streaming and batch processing of audio inputs, making it adaptable for various applications.

handy-computer / transcribe.cpp

tailwindlabs / tailwindcss

denoland / deno

angular / angular

NASA’s new dark energy space telescope can also detect killer asteroids

tailwindlabs / tailwindcss

denoland / deno

angular / angular

NASA’s new dark energy space telescope can also detect killer asteroids