Reimplementation of https://zenn.dev/cloud_ace/articles/19bd3554ac8432, but for DLsite voice works!
You have to copy existing .env.example
file to .env
file and fill in to run in local.
PINECONE_API_KEY
: Pinecone API keyPINECONE_INDEX
: Pinecone index namePINECONE_ENV
: Pod environment to host index DBOPENAI_API_KEY
: OpenAI API keyOPENAI_API_MODEL
: OpenAI modelOPENAI_API_TEMPERATURE
: OpenAI API sampling temperature (to put it simply, it is randomness)
You can fetch data from DLsite with eggplants/dojinvoice_db.
# Setup deps
poetry shell
poetry install
# Convert SQLite db to CSV file
python scripts/convert_db_to_csv.py <sqlite db file path>
# Create vector data from CSV data and upload to Pinecone
python scripts/add_doc_to_index.py <csv file path>
# Launch Streamlit server
drl