I recently started learning Chinese characters using Anki and used AwesomeTTS to generate text-to-speech MP3 files for pronunciation. I found that many characters wouldn’t get generated because of a bug however, so I took it upon myself to fix this.
I also added hash generation for the filenames based on the input sentence. This is in order to be able to mass generate files for large decks in which you’ve already generated some sentences. Because the hash generation is completely client-side you also don’t unnecessarily hammer Google’s servers (thus quickly causing a temporary ban).
* Fixed: bug causing some notes not to generate
* Changed: Filename generation (now switched to a MD5 hash generated from the input sentence). No more ridiculously long filenames!
* Added: Randomized delay between each request to Google’s servers, causing a longer generation time, but preventing being blacklisted for 24 hours
* Added: Use of Cookielib, allowing you continue generation of MP3s if you get blacklisted by going to http://translate.google.com/translate_tts?tl=zh&q=translate and correctly answering the captcha.