Transcribe Web3 lingo using Custom Vocabularies within AWS
Improve your AWS transcriptions using Custom Vocabularies.
I was transcribing a few audio recordings that contained a bunch of Web3 jargon, and I needed a better way to capture words like ERC20, ERC1155, ETH, etc. What worked for me was using AWS Translate with Custom Vocabularies.
The two examples below demonstrate a Custom Vocabulary using a table. Although it's elegant that you can create a library using the International Phonetic Alphabet (IPA) or a Phrase, it's a bummer that you have to pick one standard. You cannot have a single document with both the IPA and Phrases, nor can you use two custom library docs during a translation job. So, in other words, only pick one approach.
Create a file named CustomTableWeb3Vocabulary-IPA.txt.
Phrase IPA SoundsLike DisplayAs
git ɡ ɪ t GIT
gas ɡ æ s gas
hodl h oʊ d l HODL
mainnet m eɪ n n ɛ t mainnet
testnet t ɛ s t n ɛ t testnet
fomo f oʊ m oʊ FOMO
github ɡ ɪ t h ə b Github
node n oʊ d node
tailwind t eɪ l w ɪ n d TailwindCSS
gwei g w eɪ GWEI