Hi everyone,
I was thinking about giving users the option of a self-speaking engine alongside Windows Narrator, NVDA, and other TTS tools. Piper TTS seems perfect because it’s lightweight and offline, but I’m running into some challenges.
Right now, the only solution I came up with as a beginner is a Python script that stays running in the background and reads a shared text file, where NVGT inputs the text to the txt file, python takes the text from that file and make piper say it.
This approach works, but it has several limitations:
It’s noticeably slow compared to real-time TTS.
If the text file is updated while Piper is speaking, the interruption doesn’t work reliably the old speech might continue or get cut awkwardly.
What I’m wondering is:
Is there a way for NVGT to interact directly with Piper so that I could package everything together Piper executable, models, configs and send text in real time without relying on a shared file and Python as a middleman?
Has anyone tried a proper integration of an offline TTS engine with NVGT for a game, so that speech is fast, interruptible, and fully contained in the game package?
I should note this isn’t completely necessary for my game it just piqued my curiosity because I saw something similar in Mist World, which seems to include self-contained TTS by default. I’m mainly asking to learn and expand my knowledge of NVGT and how such integrations might work.
Any advice, examples, or alternative approaches would be really appreciated!
Thank you.