How 00mframadhani00/k.a.i.r.a-project works
K.A.I.R.A is a desktop AI assistant and VTuber-oriented chat responder. Its main workflow accepts microphone speech or YouTube live chat, builds conversation context, requests a GPT response, translates it, and produces speech and OBS subtitle files. The runtime is largely orchestrated in run.py; utility modules provide prompting, translation, speech synthesis, and subtitles. Edges reflect the shown calls and file operations. The README names additional streaming technologies, but their integration is not demonstrated in the sampled runtime, so they are not shown as connected components.
Main entry points: run.py Conversation state · promptMaker.py Prompt construction · translate.py Translation · TTS.py Speech synthesis
Components 16
Input
- Audio capture
run.py - Whisper transcription
- Live chat polling
run.py
Conversation
- Conversation state
run.py - Character identity
characterConfig/KAIRA/identity.txt - Conversation history
conversation.json - Prompt construction
utils/promptMaker.py
Response
- GPT response
- Translation
utils/translate.py
Presentation
- Speech synthesis
utils/TTS.py - Subtitle files
utils/subtitle.py - Audio playback
run.py
Other components
- Microphone user
- YouTube chat
- VoiceVox engine
- OBS display
Connections 18
- Microphone user → Audio capture: provides speech
run.py - Audio capture → Whisper transcription: submits audio
run.py - Whisper transcription → Conversation state: adds transcript
run.py - YouTube chat → Live chat polling: provides messages
run.py - Live chat polling → Conversation state: adds messages
run.py - Conversation state → Conversation history: writes history
run.py - Conversation history → Prompt construction: loads history
utils/promptMaker.py - Character identity → Prompt construction: adds identity
utils/promptMaker.py - Conversation state → Prompt construction: requests prompt
run.py - Prompt construction → GPT response: supplies context
- GPT response → Translation: passes response
- Translation → Speech synthesis: translates text
run.py - Speech synthesis → VoiceVox engine: requests synthesis
utils/TTS.py - Translation → Subtitle files: sends response
run.py - Conversation state → Subtitle files: sends question
run.py - Subtitle files → OBS display: writes display files
utils/subtitle.py - Speech synthesis → Audio playback: saves audio
utils/TTS.py - Audio playback → Microphone user: plays reply
run.py