GitHub - pvonmoradi/gsay: A simple shell script to fetch/play pronunciation of English vocabulary from Google

gsay

A simple shell script to fetch pronunciation of an English word from Google.

When you search for an Enlgish vocabulary, Google shows a pronunciation in an "answer box" at top of the page. This script fetches that hopefully human-made mp3 sound file.

Features

• Query for 2020 and 2024 sound files
• Support British and American accents
• Cache to disk (enabled by default)

Dependencies

• curl
• ffplay | mpv | pw-play : A headless mp3 player, one is enough

In a Debian-like distro, these can be installed with:

sudo apt install curl ffmpeg # or mpv

Usage

Check gsay -h:

Usage: gsay [OPTIONS] QUERY
DESCRIPTION
    A simple client to fetch/play pronounciation of words from Google

ARGUMENTS
     QUERY
        Query string
        Note: Use "" or '' if query string contains special shell characters
OPTIONS
    [--year | -y]
        Set desired database year: 2020 (default) | 2024
    [--accent | -a]
        Set desired accent: gb (default) | us
    [--link | -l]
        Only print pronounciation link. Don't play their audio
    [--no-cache | -n]
        Disable cache mode: ignore cache and don't save to cache directory
        Default: cache mode is enabled. dir: /home/username/.cache/gsay
    [--debug | -v]
        Increase log verbosity to debug
    [--version | -V]
        Print version
    [--help | -h]
        Display the help message

EXAMPLES
    gsay legend
    gsay -y 2024 -a us Leicester
    gsay -n -y 2020 -a gb Forte
    gsay Pneumonoultramicroscopicsilicovolcanoconiosis
    gsay -l perfect
    gsay carte blanche

Notes

• I may be wrong but the 2024/04/19 pronounciations sound synthetic to me! Hence, 2020/04/29 is default despite being slower and less exhaustive. Caller scripts can run it like gsay -y 2020 || gsay -y 2024
• The HTTP URLs are faster, HTTPS ones are provided as comments too
• Fun examples: echo Supercalifragilisticexpialidocious Antidisestablishmentarianism Grandiloquent | xargs -n1 gsay
• Oxford 3000 word list pronunciations, limited to first 1000 words, can be downloaded (cached) to disk like this:

curl -sL https://github.com/sapbmw/The-Oxford-3000/raw/refs/heads/master/The_Oxford_3000.txt \
    | tr -d '\r' | head -1000 | xargs -I {} gsay -l "{}"

Development

• The early versions of this script supported a scraping mode besides the current heuristic mode. But, recently Google has been preventing scrapers from accessing its search results page. I tested their "Custom Search JSON API" from "Programmable Search Engine", but it doesn't provide the output from Answer Box that contains pronounciations
• linter: shellcheck
• formatter: shfmt -i 4 -bn -ci -sr