AssemblyAI CLI

Independent JSON-first CLI over the AssemblyAI speech-to-text API. Not an official AssemblyAI product. Plain fetch, Node >= 20.

cd api-apps/assemblyai-api && npm install --workspaces=false && bun run build
node dist/index.js --help

Auth

API key from https://www.assemblyai.com/dashboard/api-keys.

<secret-manager> | assemblyai-api auth --key-stdin    # validates, saves ~/.config/assemblyai-api/credentials.json (0600)
# or: ASSEMBLYAI_API_KEY=... assemblyai-api whoami
# EU data zone: ASSEMBLYAI_REGION=eu

Commands

| command | endpoint | |---|---| | auth, whoami, logout | GET /v2/transcript?limit=1 (no account endpoint exists) | | upload <file> | POST /v2/upload | | transcribe <url\|file> [--speaker-labels] [--language-code] [--data json] [--wait] | POST /v2/transcript (+ polling) | | get <id> [--wait], list, delete <id> | /v2/transcript[/{id}] | | sentences, paragraphs, redacted-audio, subtitles <id> srt\|vtt, word-search <id> <words> | /v2/transcript/{id}/... | | chat --model m --prompt p, models | LLM Gateway https://llm-gateway.assemblyai.com/v1 (paid plans) | | streaming-token --expires-in-seconds n | GET https://streaming.assemblyai.com/v3/token | | api <METHOD> <path> [--host api\|llm\|streaming] | anything else |

Real-time streaming itself is a WebSocket (wss://streaming.assemblyai.com/v3/ws); use the official SDK for that.