Oído: Open-vocabulary speech recognition on a $5 ESP32-S3 (3.7% LibriSpeech WER) | Hacker News - But Better

Oído: Open-vocabulary speech recognition on a $5 ESP32-S3 (3.7% LibriSpeech WER)github.com

Oído: open-vocabulary speech recognition that fits in a $5 ESP32-S3. 3.7% LibriSpeech WER, no cloud, no NPU. - lokutor-ai/oido

2 points•dani-lokutor•about 14 hours ago•0 comments•

#speech #vocabulary #esp32 #recognition #librispeech #wer #open vocabulary #vocabulary speech

0 comments

No comments yet.

Read the full thread on Hacker News →

Related stories

You Can’t Stay Here: The Efficacy of Reddit’s 2015 Ban Examined Through Hate Speechcomp.social.gatech.edu
Lobsters · 12 points · about 9 years ago
Jinfer-Parakeet: Automatic Speech Recognition for the JVMgithub.com
Hacker News · 1 points · 6 days ago
Parakeet.java: Automatic Speech Recognition in Pure Javagithub.com
Hacker News · 1 points · 7 days ago
Building a 512-LED wall that dances to my music: FFT, ESP32, and a custom serial protocoldev.to
DEV Community · 5 points · 8 days ago
I put Spotify on a $25 Win98 desk player (ESP32, no PSRAM)dev.to
DEV Community · 0 points · 4 days ago
Running Linux with kernel 6.3 on Open Source Hardware board with ESP32-S3olimex.wordpress.com
Lobsters · 8 points · over 3 years ago