TwAIlight operations manual

Demonstration video - https://drive.google.com/open?id=1nvMCFL9rfiDm3sO9PJIW8gqYy833fnFX

Step 1 - Download “win_executable.exe” from https://github.com/Twibot-ai/pony_synth_script/releases

Step 2 - Download “20180505_deepvoice3_ljspeech.json” and “checkpoint_step000200000_22khz_16bit.pth” from https://drive.google.com/drive/u/1/folders/1Lej38TiSpg9vB11gTwaKTgiOE_pT6Dko

Step 3 - Place the three things you downloaded into one folder, then hold shift and right-click in the file window and select “Open PowerShell window here”.

Step 4 - Copy and paste the following into the command window:
.\win_executable.exe "./checkpoint_step000200000_22khz_16bit.pth" "./20180505_deepvoice3_ljspeech.json" "This is a test." "test.wav"

Step 5 - Press the enter key to run the command and wait for the program to do it’s thing (~10 - 20 seconds). It will output the sound file “test.wav” to the same folder the programme files are stored in.

Step 6 - Replace the "This is a test." and "test.wav" text with the words you want to generate, and the file name you want respectively. Make sure to include the “ marks around each part and .wav for the file name.

Example:
.\win_executable.exe "./checkpoint_step000200000_22khz_16bit.pth" "./20180505_deepvoice3_ljspeech.json" "I want ear scratchies." "Ear scratchies.wav"

Right now, TwAIlight can’t generate more than about 10 seconds of audio in one go, so it’s generally best to only do one sentence at a time.

Step 7 - Poni poni poni.