blog post: [https://www.bryanthornbury.com/posts/fast-forced-alignment](https://www.bryanthornbury.com/posts/fast-forced-alignment)
github: [https://github.com/brthor/fast-forced-aligner](https://github.com/brthor/fast-forced-aligner)
I developed optimized CUDA kernels to speed up my forced alignment pipelines using Astra.
To my surprise, I was able to achieve speeds 10x - 1000x faster than popular existing open source options, including the torchaudio implementations I was using before.
I'm sharing them for anyone else who finds them useful.
The acceleration from AI is getting wild.
Comments URL: [https://news.ycombinator.com/item?id=50310134](https://news.ycombinator.com/item?id=50310134)
Points: 1
source & further reading
github.com — original article
ProgramBench: Can Language Models Rebuild Programs from Scratch?
Show HN: Automatic deduction of FlashAttention kernel from ~25 lines of SQL
EasyRE-Agent – an AI agent that solves crackmes and checks its own answers