Configuration

All settings are loaded from environment variables (or a .env file) using the AUDIA_ prefix. Run audia info at any time to inspect the active configuration.

cp .env.example .env   # create your .env from the template
audia info             # verify active settings

LLM provider (required)

LLM curation is mandatory — audia cannot convert documents without it.

Option A — OpenAI

AUDIA_LLM_PROVIDER=openai
AUDIA_OPENAI_API_KEY=sk-...
AUDIA_LLM_MODEL=gpt-4o-mini           # default; gpt-4o for higher quality

Custom endpoint (Azure OpenAI, corporate proxy, or any OpenAI-compatible URL):

AUDIA_OPENAI_API_BASE=https://your-org.openai.azure.com/

Option B — Anthropic

AUDIA_LLM_PROVIDER=anthropic
AUDIA_ANTHROPIC_API_KEY=sk-ant-...
AUDIA_LLM_MODEL=claude-3-5-haiku-20241022

Custom endpoint:

AUDIA_ANTHROPIC_API_BASE=https://your-proxy.example.com/

Shared LLM settings

Variable	Default	Description
`AUDIA_LLM_MAX_CHUNK_CHARS`	`8000`	Characters per LLM curation chunk
`AUDIA_LLM_TEMPERATURE`	`0.1`	0 = deterministic, 1 = creative

TTS backend

Variable	Default	Description
`AUDIA_TTS_BACKEND`	`edge-tts`	`edge-tts` \| `kokoro` \| `openai`
`AUDIA_TTS_VOICE`	`en-US-AriaNeural`	Voice name (backend-specific)
`AUDIA_TTS_RATE`	`+0%`	edge-tts speaking rate, e.g. `+10%`
`AUDIA_TTS_CHUNK_CHARS`	`3800`	Characters per TTS chunk

edge-tts voices (no key required):

AUDIA_TTS_VOICE=en-US-AriaNeural   # female, US
AUDIA_TTS_VOICE=en-US-GuyNeural    # male, US
AUDIA_TTS_VOICE=en-GB-SoniaNeural  # female, British

List all available voices: edge-tts --list-voices

Kokoro (pip install audia[kokoro]):

AUDIA_TTS_BACKEND=kokoro
AUDIA_TTS_VOICE=af_heart

OpenAI TTS:

AUDIA_TTS_BACKEND=openai
AUDIA_TTS_VOICE=nova    # alloy | echo | nova | shimmer | onyx | fable

STT (voice input)

Requires: pip install audia[stt] (included in core dependencies).

Variable	Default	Description
`AUDIA_STT_MODEL`	`base`	`tiny` \| `base` \| `small` \| `medium` \| `large-v3`
`AUDIA_STT_DEVICE`	`cpu`	`cpu` \| `cuda`
`AUDIA_STT_RECORD_SECONDS`	`30`	Max microphone recording duration

Storage

Variable	Default	Description
`AUDIA_DATA_DIR`	`~/.audia`	Root directory for DB, audio, uploads, debug

~/.audia/
  audia.db          ← SQLite database
  audio/            ← generated .mp3 files
  uploads/          ← uploaded PDFs
  debug/            ← per-run debug text snapshots

Web server

Variable	Default	Description
`AUDIA_SERVER_HOST`	`127.0.0.1`	FastAPI bind host
`AUDIA_SERVER_PORT`	`8000`	FastAPI bind port
`AUDIA_RELOAD`	`false`	Uvicorn auto-reload (dev only)

Research

Variable	Default	Description
`AUDIA_ARXIV_MAX_RESULTS`	`10`	Max papers returned per ArXiv query

Full `.env.example`

See the annotated .env.example for a complete template.