This project was made to help out Patricia Ryter for her matura project.
She will experiment if it is possible to learn a language
while sleeping.
For that, the spoken word recordings are taken from Quizlet.
The application should generate an audio file with all the words of the learn-set. They have to be separated in different blocks of 20 word pairs (base lang + translation) which will play in a loop 18 times.
Quizlet doesn't provide the audio in their API but maps the audio in the website,
so the program first downloads the learn-set site HTML.
Then it extracts the links to the audio files and downloads them.
Note: this project last was used in 2021 and might probably require some adjustments to work with the current Quizlet website.
Silences are created and then the audio is concatenated using ffmpeg demuxer.
Some of the audio Files are hosted externally on an amazon server.
They have a different sample rate (44'100Hz), so they have to be downloaded
and converted to the same sample rate than those hosted on Quizlet (32'000).
The speed is approximately 1s per 2 words (base + translation) if the sample rate doesn't have to be converted, but this varies greatly depending on the server performance.
A control-file which is basically all words spoken one after the other without redundancy is also provided to check if all the words are spoken as expected.