English Words Pronunciation MP3 Audio Download

Introduction

This is a simple Python script to download the MP3 pronunciation audio for 119,376 unique English words/terms. They will take around 2 GB on your disk in total.

The MP3 sources were obtained from the following 7 online dictionaries:

This project should have covered everything you might ever think of, from ".22 caliber" to "zinc cadmium sulfide"; from "1000000000000" to "level dependent functional magnetic resonance imaging" (YES, we have the pronunciation for this long term). If the word/term you are looking for is not in the list, it's most likely because the it didn't appear in any of the online dictionaries.

I currently don't have the intention to share the source code of my crawler framework that found these MP3 URLs. However, if there are more sites you wanna crawl but you don't know how to, ask me for help and I'll happily make a crawler for you.

Getting Started

Download & install Python 3
Install dependencies

$ pip install -r requirements.txt

Download All MP3

Just do it.

$ python3 download_all_mp3.py

This command will start a downloader with 30 threads by default.

To specify how many threads you want to use in the downloader, just pass in a positive integer as parameter:

$ python3 download_all_mp3.py 10

This will start a downloader with 10 concurrent threads.

The downloaded mp3 audio files will be stored in the download/ directory

Use `data.json` and `ultimate.json` in Other Applications

The data.json and ultimate.json file contains the URLs to the pronunciation MP3 audio files for 119,376 unique English words.

My own crawler framework took tens of hours to get all the data from the Web. Now, you can use the data directly free of charge, instead of having to spend your time and effort to write the crawler and to crawl the Web. The idea behind this project is just like rainbow tables.

The only difference between data.json and ultimate.json is, the former contains only one URL for each word/term, while the latter have all the URLs that ever appeared in all online dictionaries.

For example, a segment of the data.json file looks like:

{
	"abel": "http://static.sfdict.com/staticrep/dictaudio/A00/A0015900.mp3",
	"abele": "http://www.yourdictionary.com/audio/a/ab/abele.mp3",
	"abelia": "http://s3.amazonaws.com/audio.vocabulary.com/1.0/us/A/1IFDVKNEVQTHP.mp3",
	"abelmoschus": "http://img2.tfd.com/pron/mp3/en/US/sh/shdjdgdyslsjsfsodrsksns3h7h3.mp3",
	"abelmosk": "http://img2.tfd.com/pron/mp3/en/UK/dr/drdjdgdyslsjsfsoshgk.mp3",
	"abenaki": "http://img2.tfd.com/pron/mp3/en/US/dt/dtdjdgdysjddskslhn.mp3",
	"aberdare": "http://img2.tfd.com/pron/mp3/en/UK/df/dfdjdgdysddtdsstd7gk.mp3",
	"aberdeen": "http://www.yourdictionary.com/audio/a/ab/aberdeen.mp3"
}

While the corresponding segment in ultimate.json:

	"abel": [
		"http://static.sfdict.com/staticrep/dictaudio/A00/A0015900.mp3",
		"http://img2.tfd.com/pron/mp3/en/US/d5/d5djdgdyslht.mp3",
		"http://img2.tfd.com/pron/mp3/en/UK/d5/d5djdgdyslht.mp3",
		"http://www.yourdictionary.com/audio/a/ab/abel.mp3"
	],
	"abele": [
		"http://www.yourdictionary.com/audio/a/ab/abele.mp3",
		"http://static.sfdict.com/staticrep/dictaudio/A00/A0016300.mp3",
		"http://www.oxforddictionaries.com/media/english/uk_pron/a/abe/abele/abele__gb_2_8.mp3",
		"http://s3.amazonaws.com/audio.vocabulary.com/1.0/us/A/1B3JGI7ALNB2K.mp3",
		"http://www.oxforddictionaries.com/media/english/uk_pron/a/abe/abele/abele__gb_1_8.mp3"
	],
	"abelia": [
		"http://s3.amazonaws.com/audio.vocabulary.com/1.0/us/A/1IFDVKNEVQTHP.mp3",
		"http://www.oxforddictionaries.com/media/english/uk_pron/a/abe/abeli/abelia__gb_1_8.mp3",
		"http://www.yourdictionary.com/audio/a/ab/abelia.mp3",
		"http://static.sfdict.com/staticrep/dictaudio/A00/A0016400.mp3"
	],
	"abelmoschus": [
		"http://img2.tfd.com/pron/mp3/en/US/sh/shdjdgdyslsjsfsodrsksns3h7h3.mp3"
	],
	"abelmosk": [
		"http://img2.tfd.com/pron/mp3/en/UK/dr/drdjdgdyslsjsfsoshgk.mp3",
		"http://static.sfdict.com/staticrep/dictaudio/A00/A0016700.mp3",
		"http://s3.amazonaws.com/audio.vocabulary.com/1.0/us/A/G9XTJLHSNJWL.mp3"
	],
	"abenaki": [
		"http://img2.tfd.com/pron/mp3/en/US/dt/dtdjdgdysjddskslhn.mp3",
		"http://static.sfdict.com/staticrep/dictaudio/A00/A0016800.mp3",
		"http://www.yourdictionary.com/audio/a/ab/abenaki.mp3"
	],
	"aberdare": [
		"http://img2.tfd.com/pron/mp3/en/UK/df/dfdjdgdysddtdsstd7gk.mp3",
		"http://img2.tfd.com/pron/mp3/en/US/df/dfdjdgdysddtdsstd7gk.mp3",
		"http://static.sfdict.com/staticrep/dictaudio/A00/A0017100.mp3"
	],
	"aberdeen": [
		"http://www.yourdictionary.com/audio/a/ab/aberdeen.mp3",
		"http://static.sfdict.com/staticrep/dictaudio/A00/A0017200.mp3"
	]

Apparently, each key is an English word and the value is either the URL, or a list of URLs, to the MP3 audio files.

The data.json and ultimate.json are 11.1 MB and 39.1 MB respectively, so you can load any of them into the memory and let it serve as a look-up table.

Acknowledgements

The list of words is downloaded from a third-party distributor of WordNet called Words API Blog.

My special thanks to these online dictionary owners for making their websites crawlable. Some of them apparently devoted quite some effort hiding out the MP3 URls from the main HTML files though.

thegithubr / english-words-pronunciation-mp3-audio-download Goto Github PK