discord bot for texting, talking, and creating covers with favourite characters using retrieval-based voice conversion.
- download RVC-WebUI Project
- change directory to mibano and create virtual environment
cd ~/mibanopython3.10 -m venv venvsource venv/bin/activate
- download your version of pytorch
pip install -r requirements.txt- add environment variables to .env (OPENAI_API_KEY="", DISCORD_TOKEN="", RVC-PATH="rvc-folder-name",)
- add baseline.pth to mibano/vocal-remover/models
python bot-main.py
additional note:
- may need to add baseline.pth to vocal-remover assets
- may need to reference an index for rvc rendering
- change path in cover generation
general commands for chatting with kenoibob!
?help- show this message
?speaking- show current speaker
?switch <speaker>- switch speaker from: {valid_speakers}
?say <text>- uses current voice to speak
?ai <text>- generates response and speaks it with current voice
create covers of YouTube videos using voice conversion.
?cover <speaker> <pitch> <start_time> <youtube_url>- cover a song using voice conversion
?advcover <speaker> <pitch> <gain> <reverb> <start_time> <youtube_url>- ?cover but with more options
Parameters
variables for tweaking settings of covers.
speaker = {valid_speakers} (string).` Specifies speaker to sing with
pitch = (integer -72 to ~72).` Specifies pitch in semitones to sing with
gain = (float, 0 to ~30).` Specifies gain adjust in dB of singing
reverb = (float 0 to 100).` Specifies reverb % mix
start_time = (integer 0 to duration-45s).` Specifies start time in seconds of song
broken code, use at own risk.
?talk- start listening to your voice channel and talks to you in the current voice unfinished
?stop- stops ?talk unfinished