*About This Video*
The input and output were done by text, exactly as shown.
I attempted to be as spontaneous as Meta Llama 3
(LLaMMa3:8b-instruct-q8_0),
responding to its questions with what I could think of and make up on the moment,
then building on that during our conversation.
*Technical Details*
The model used was Meta Llama 3 (LLaMMa 3:8b-instruct-q8_0)
run locally using ollama.
For voice synthesis, I utilized
PIPER TTS from GitHub github.com/rhasspy/piper
My voice was set to
en_US-joe-medium,
while LLaMMa's voice was set to
en_US-ljspeech-high.
*Audio and Video Capture*
I used audio vizualization scope, courtesy of the VLC player,
captured using OBS Studio.
Additional software used includes:
* DaVinci Resolve
* Audacity
* Notepad++
* FFmpeg
* FFprobe
Thanks again for watching!