Week 4 Progress Report

Google Summer of Code
1

Hi,
This is my week 4 progress report.

Works Done:

  1. Improved on my previous work of updating the phonetic dictionary. The previous script was not working due to the file format of the unzipped audio files not being .wav but .mp3 and some other errors. Solved them.
    https://github.com/AnirbanBanik1998/Modern_Speak_and_Spell/commit/5002451f12eb3f08a851eb1158e149199c7c0620
    https://github.com/AnirbanBanik1998/Modern_Speak_and_Spell/commit/add538c483a5de7a53b06c5a8da25903ceff6cac

  2. Updated the speech processing of combined audio files
    https://github.com/AnirbanBanik1998/Modern_Speak_and_Spell/commit/17df84ad856ccdddd7d8e2a098a557fa80dc74be

Also Updated the dictionary based on the results of the live recording from the microphone.
https://github.com/AnirbanBanik1998/Modern_Speak_and_Spell/commit/7e2699583981de369e8ea57fe37e04fdfa09dd22

  1. Worked on automating the creation of acoustic model for each new user.
    https://github.com/AnirbanBanik1998/Modern_Speak_and_Spell/commit/204cdd27235d50bce2627359fbaa149e92c326ac

  2. Started working on a generalized recording and decoding script to improve reusability and reduce code repeats.
    https://github.com/AnirbanBanik1998/Modern_Speak_and_Spell/commit/c42a77c847cc538838eec7c40e7a20b3dac23e9f

Issues faced:

  1. For the training of the acoustic model for each new user to adapt to his/her accent…it requires him/her to speak the transcribed words given…for making the files containing the recorded voice required for the training. It can be quite time-consuming and tiresome for the user. So maybe thinking of scraping this idea and now going forward with the default acoustic model. Subjected to feedback.

Works to be done:

  1. Finishing work on the generalized recording and decoding scripts, and using them in all the games and also the launcher.

  2. Further work on speech-processing.

  3. Starting work on the game-graphics.

Regards,

Anirban