Almost none of Africa's 2,000+ languages have working ASR. Fine-tuned Whisper on a cleaned 144h Kikuyu corpus. First model: 32.9% WER / 10.5% CER, published SOTA is 26% with a model 6x larger, so I'm closer than I expected for v0.
Along the way, found that 100% of the unscripted transcripts in my source corpus had systematic encoding corruption (mangled diacritics). Would love tips/stories from anyone who's done low-resource ASR.
Almost none of Africa's 2,000+ languages have working ASR. I'm fine-tuning Whisper on ~217h of transcribed Kikuyu (no Kikuyu token exists, so I'm repurposing the Swahili one). A recent paper hit ~26% WER with similar data, so a usable v0 is realistic.
2nd coolest website on the internet
https://github.com/thirtyninetythree/gtav-leetcode-extension
Hey HN! I forked an fun little extension that plays GTA mission failed/passed sound and displays banner based on Leetcode solution submissions.Try it out and have a laugh in the grind!
https://github.com/thirtyninetythree/stealthislook
A better way to shop. Take a photo and search online for the clothes