kombify SpeechKit

SpeechKit Server Setup

Guided onboarding for provider, model, and credential settings on this server instance.

Smoke

SpeechKit can run fully self-hosted out of the box. The default stack uses Whisper Large v3 Turbo for dictation and a local llama.cpp model for Assist and the Voice Agent pipeline.

Default local LLM: ggml-org/gemma-4-E2B-it-GGUF:Q8_0

Runtime waiting -

Models waiting -

Data waiting -

Choose provider types and concrete models for Dictation, Assist, and Voice Agent.

Onboarding pending

Dictation

Provider type Provider Model

Assist

Provider type Provider Model

Voice Agent

Provider type Provider Model

Dictation Dictionary

Custom words

Assist Tools

Voice Agent Prompt Template

Agent profile System prompt

Enable admin sign-in Password required

Create the admin account for this SpeechKit setup UI. API clients still use their configured bearer token.

Admin username Admin password Confirm admin password

Generate server API token Token pending

Keep this on to generate a bearer token for Windows clients and API callers. Turn it off when authentication is handled outside SpeechKit or the API should stay available with the current server auth mode.

Bearer token env

Generated server API token - Store this token in the Windows client environment and send it as Authorization: Bearer.

OpenAI Env API Key

Groq Env API Key

Google Env API Key

Hugging Face Env API Key

OpenRouter Env API Key

Review the active choices. Saving marks onboarding complete for this deploy version and writes the desired model settings.

Dictation - -

Assist - -

Voice Agent - -

API Auth - -