Voice Mode (STT/TTS)

Talk to Yukkai using voice input and hear responses spoken aloud.

Overview

Voice Mode lets you interact with Yukkai entirely by speech. Instead of typing your messages, you speak naturally — and instead of reading the response on screen, Yukkai reads it back to you aloud.

Voice Mode combines two complementary technologies:

Technology	Abbreviation	Role
Speech-to-Text	STT	Transcribes what you say into text Yukkai can process
Text-to-Speech	TTS	Converts Yukkai's text response into spoken audio

This makes Voice Mode a fully conversational experience. You talk, Yukkai listens and replies out loud, and the cycle continues — just like a phone call with a knowledgeable assistant.

Voice Mode is great for:

Hands-free use — while cooking, driving, or walking.
Accessibility — if typing is difficult or impractical.
Multi-tasking — talk to Yukkai while your screen is occupied.
Natural conversation — some questions are simply easier to ask out loud.

How It Works

The voice pipeline flows in a straight line:

🎤 Microphone → 📝 STT (Speech Recognition) → 💬 Conversation Engine → 🔊 TTS (Speech Synthesis) → 🔈 Speaker

1. Speech-to-Text (STT)

When Voice Mode is active, Yukkai listens to your microphone. The audio is transcribed into text in real time using a speech recognition engine.

On-device recognition — uses the native macOS Speech Framework

(SFSpeechRecognizer). It supports multiple languages (including French and English) and provides partial results as you speak, so you can see the transcription live.

Advanced recognition (optional) — an alternative provider such as OpenAI

Whisper can deliver higher accuracy, especially for technical terms, proper nouns, or code snippets. This may run through an API or a local model depending on your configuration.

2. Conversation Engine

Once your speech has been transcribed, the resulting text is fed into Yukkai's normal conversation engine. Yukkai processes it exactly as it would a typed message — it can answer questions, run tools, search the web, manage tasks, and more.

3. Text-to-Speech (TTS)

When Yukkai has a response ready, the text is converted into spoken audio via speech synthesis.

Native synthesis — uses the built-in macOS AVSpeechSynthesizer or an

external provider depending on your settings.

Configurable output — you can adjust the voice, speaking rate, pitch, and

volume.

Queued playback — long responses are queued so you hear the full answer

without gaps or truncation.

Activation Modes

Yukkai supports two ways to trigger voice input:

Mode	How it works	Best for
Push-to-Talk	Hold a keyboard shortcut while you speak. Release when you're done — your message is sent automatically.	Noisy environments, precise control, quick questions
Hands-free (Voice-Activated)	Yukkai uses Voice Activity Detection (VAD) to detect when you start and stop speaking. No keyboard or mouse needed.	Natural conversation, accessibility, hands-free scenarios

In hands-free mode, a configurable silence threshold determines how long Yukkai waits after you stop talking before it treats your message as complete. This lets you pause mid-sentence without the system cutting you off.

How to Use It

Prerequisites

Before you start, make sure the necessary permissions are granted:

Microphone — open System Settings → Privacy & Security → Microphone

and enable Yukkai.

Speech Recognition — open System Settings → Privacy & Security →

Speech Recognition and enable Yukkai.

Input device — confirm the correct microphone is selected in macOS

sound settings.

If Yukkai doesn't seem to hear you, close other apps that may hold exclusive microphone access, then relaunch Yukkai.

Starting Voice Mode

Activate Voice Mode via the app interface or the designated keyboard

shortcut.

A visual recording indicator appears, confirming that Yukkai is

listening.

Start speaking. If you're in push-to-talk mode, hold the shortcut while you

talk and release when you're done. In hands-free mode, just talk naturally.

Yukkai transcribes your speech in real time, processes your request, and

reads the response aloud.

Stopping Voice Mode

Toggle Voice Mode off via the same shortcut or interface control.
In hands-free mode, you can often simply say a stop phrase or let silence

time out, depending on your configuration.

Configuration Options

Setting	Description
Recognition language	Choose the language Yukkai listens for (e.g., English, French).
STT provider	On-device speech framework or an advanced provider like Whisper.
TTS voice	Select from available system or provider voices.
Speaking rate	Adjust how fast Yukkai reads responses aloud.
Pitch & volume	Fine-tune the audio character of the synthesized voice.
Silence threshold	In hands-free mode, how long to wait after speech ends before sending.
Keyboard shortcut	The key combination for push-to-talk activation.

Examples

Example 1 — Quick Question (Push-to-Talk)

You (hold shortcut): "What's the weather like in Paris today?"

(release shortcut)

> Yukkai (spoken): "Right now in Paris it's 18 degrees Celsius with

scattered clouds. Expect light rain this afternoon around 3 p.m."

Example 2 — Hands-Free Conversation

You: "Add a task to review the pull requests before Friday."

> Yukkai (spoken): "Done. I've created a task titled 'Review pull

requests' with a due date of this Friday. Would you like me to set a

reminder?"

> You: "Yes, remind me Thursday morning."

> Yukkai (spoken): "I'll remind you on Thursday at 9 a.m."

Example 3 — Complex Multi-Step Request

You: "Search YouTube for the latest WWDC keynote and summarize the main

announcements."

> Yukkai (spoken): "I found the WWDC 2025 keynote. The main announcements

include a new design language for iOS, on-device AI improvements, and a

redesigned developer toolkit. Would you like more detail on any of these?"

In every case, Yukkai processes your spoken words just as it would typed text — so all your favourite tools and capabilities remain available in Voice Mode.

Tips

Improving Recognition Accuracy

Reduce background noise. Recognition works best in a quiet environment.

If accuracy is poor, try moving to a quieter room or using a headset microphone.

Speak clearly at a normal pace. You don't need to slow down dramatically,

but avoid mumbling or speaking too quickly.

Test with a short phrase. If something seems off, try a simple sentence

like "Hello, can you hear me?" to verify that recognition is working.

Choose the right language. Make sure the recognition language matches the

language you're speaking. Switching mid-sentence can confuse the recognizer.

Try an alternative provider. If the on-device recognizer struggles with

technical jargon or accents, switching to a provider like Whisper may improve results.

Getting the Best TTS Experience

Pick a natural voice. macOS offers several voices — some sound more

natural than others. Try a few and pick the one you prefer.

Adjust the speaking rate. The default rate may feel too fast or too slow.

Tweak it until Yukkai's responses sound comfortable.

Use headphones for privacy. If you're in a shared space, headphones

prevent Yukkai's responses from being heard by others.

Choosing an Activation Mode

Use push-to-talk in noisy environments or when you want precise control

over exactly what gets sent. It avoids accidental activation from background conversation.

Use hands-free when you want a natural, flowing conversation — for

example, during a walk or while cooking. Adjust the silence threshold so Yukkai doesn't cut you off mid-sentence or wait too long before responding.

Accessibility

Voice Mode is particularly valuable for users who find typing difficult or impractical. The hands-free mode enables a fully hands-free conversational experience, and the TTS output means you never need to look at the screen to get your answer.

Troubleshooting Checklist

If Voice Mode isn't working as expected, work through these steps:

Check Speech Recognition permission in System Settings.
Check Microphone permission in System Settings.
Confirm the selected recognition language.
Test with a short, clear phrase.
Reduce background noise.
Confirm the correct input device is selected.
Close apps that may hold exclusive microphone access.
Relaunch Yukkai.
Try an alternative speech provider or on-device mode.

If issues persist after all these steps, the problem may be with the selected speech service itself — check for service outages or configuration errors.