Talk to your AI
naturally.

Whisper-powered speech recognition, natural text-to-speech responses, and hands-free operation. Have a real conversation with your AI — no keyboard required.

Try Voice Interface Read the docs

What makes it powerful

Whisper STT

OpenAI's Whisper model runs locally for speech-to-text. Accurate transcription in 99 languages with near-zero latency.

Natural TTS responses

AI responses are spoken aloud with natural intonation. Choose from multiple voice profiles or clone your preferred voice.

99 language support

Speak in your native language and get responses back in any language. Real-time translation built into the conversation flow.

Wake word activation

Say "Hey Alabobai" to start a conversation hands-free. Customizable wake word with local-only detection — no always-on cloud mic.

Conversation memory

Voice conversations maintain full context. Reference earlier topics, ask follow-ups, and build on previous exchanges naturally.

Private by design

All voice processing happens on-device. Your voice data is never sent to external servers. Transcripts are encrypted locally.

See it in action

A natural conversation with your AI, entirely hands-free.

Wake word detected: "Hey Alabobai"

Listening...

You: "Summarize the last three research papers I indexed about quantum computing."

Transcribed in 0.3s (Whisper large-v3, local)

Querying Local AI Brain... 3 papers found

Alabobai: "The three papers cover complementary aspects of quantum error correction. The first, by Chen et al., proposes a new surface code variant that reduces physical qubit overhead by 40%. The second paper demonstrates a practical implementation of logical qubit operations on IBM's 127-qubit processor. The third introduces a novel decoding algorithm that runs in linear time..."

Response spoken via TTS (Piper, natural voice)

Ready for follow-up

How it works

Speak naturally

Use the wake word or press a hotkey to start. Whisper transcribes your speech locally with high accuracy in any language.

AI processes & responds

Your query is processed by your local LLM with full access to your knowledge base, tools, and conversation history.

Hear the answer

The response is spoken aloud with natural TTS. Continue the conversation, ask follow-ups, or switch to text at any time.

Technical specifications

STT engineWhisper (large-v3)

TTS enginePiper / Coqui

Languages supported99 (STT) / 30+ (TTS)

Transcription latency<500ms (GPU)

Wake word detectionLocal, <2ms

Voice profiles12 built-in + custom

Audio processing100% on-device

Data stored in cloudNone

Talk to your AInaturally.