Automatic Speech Recognition Class
Taught by Joseph Tepperman
Meeting times —
Thursday June 3, 7-10pm
Sunday June 6, 1-4pm
Thursday June 10, 7-10pm
Sunday June 13, 1-4pm
Become a Machine Project Member and get a discount on all classes for a year!
Learn how to get a computer to respond to your speech!
This class is an introduction to Automatic Speech Recognition, for people who want to use it in music or visual art or anything.
All the software we’ll use is free and open-source, and you can keep all the code and notes from the class.
Here’s what we’ll cover:
Class 1 - basic Speech Recognition theory; software installation and tutorial
Class 2 - ordinary recognizer use: alignment, decoding, acoustic model training, pronunciation scoring
Class 3 - custom recognition: adaptation, modeling pronunciation and grammar, etc.
Class 4 - extensions to art and music
You’ll need to bring your own laptop, Mac or Linux preferred. You’ll also need a cheap headset microphone.
There won’t be any serious programming, but you should be able to code some basic things, and you need to be familiar with running programs from the command line.
It helps if you already know a little about probability theory and phonetics, since we’ll be talking about that stuff a lot. But you don’t have to be an expert to make your recognizer work (or to have fun).
About the instructor:
Here you can see/hear some speech-art installations he’s made, and also read his research publications.
images by Dorian Wood
Please note, all class fees include a non-refundable enrollment deposit of $25 that will be deducted from your refund if you sign up for, then drop, a class. So, for example, if you sign up for Machine Sewing 101 and pay the $155 class fee, but then remember that you have trapeze school final exams that conflict with the Sewing class and shouldn’t have signed up after all, we will refund you $130 of your tuition payment.