ARMES Voice: Hear Your AI Think Out Loud

Sometimes the best way to absorb an answer is to hear it.

You are in the middle of cooking and asked your AI to walk you through a recipe. You are on a long drive and want to catch up on a research summary you requested earlier. You are reviewing a draft and want to hear how it actually sounds out loud before you send it. Or you simply learn better by listening.

AI Voice is live on ARMES. Tap the speaker icon on any AI response and hear it read aloud — with the same zero-data-retention privacy you rely on for every other interaction. Your text is processed and immediately forgotten. Never stored, never profiled, never used for training.

How it works

Every AI response in ARMES now has a small speaker icon alongside the existing Copy, Save, and Feedback buttons. Tap it, and ARMES converts that response into spoken audio using a neural text-to-speech model.

That is it. No setup required, no separate app, no copy-pasting into another tool.

The speaker button is available on both of ARMES' chat surfaces:

• The Chat page — your primary conversation view
• The Command Center's chat tab — the quick-access panel in your workspace

When audio is playing, an inline playback bar appears right inside the message bubble. You can pause, resume, scrub through the audio, or restart from the beginning. Tap the speaker icon again to pause; tap it once more to resume. If you start speaking a different message, the current one stops automatically — only one message plays at a time.

Choose your voice

Head to Settings > Voice (or the Voice tab in the Command Center's settings panel) and you will find everything you need to make ARMES sound the way you want.

Pick a voice

We are launching with 12 distinct neural voices, each with its own character:

Voice	Character
Leda (default)	Youthful and bright
Kore	Firm and confident
Puck	Upbeat and lively
Zephyr	Bright and warm
Charon	Informative and steady
Fenrir	Excitable and dynamic
Aoede	Breezy and natural
Orus	Firm and grounded
Callirrhoe	Easy-going and smooth
Autonoe	Bright and clear
Enceladus	Breathy and soft
Umbriel	Easy-going and calm

Every voice has a play button right next to it in Settings, so you can audition each one before committing. Tap play, listen for a few seconds, and pick the one that fits.

Set your speed

A speed slider lets you dial playback anywhere from 0.5x (half speed — great for absorbing dense material) to 2.0x (double speed — for skimming or when you just want the gist fast). The default is 1.0x. The slider applies to both manual reads and auto-read.

Auto-read mode

If you want ARMES to speak every new response as it arrives — hands-free, no tapping required — flip on Auto-read responses. New agent answers will be read aloud automatically. Turn it off, and you are back to on-demand with the speaker button.

When to use it

AI Voice is not just a convenience feature. Here are some of the ways it changes how you work with your AI:

• Multitasking. Listen to a response while you cook, drive, exercise, or do anything that keeps your hands busy.
• Proofreading. Hearing text out loud catches awkward phrasing, run-on sentences, and tone mismatches your eyes skip over. Ask your AI to draft something, then listen to it before you use it.
• Learning and retention. Research shows that hearing information improves comprehension and recall. If you use ARMES for study, research, or exploration, listening to answers gives your brain a second pass.
• Accessibility. For anyone who finds reading on screens difficult — whether due to vision, attention, or fatigue — voice turns ARMES into something you can use with your eyes closed.
• Long responses. When an AI gives you a thorough, multi-paragraph answer, sometimes it is easier to lean back and listen than to scroll and read.

Privacy: always zero data retention

AI Voice runs through the same zero-data-retention infrastructure as every other AI interaction on ARMES. Your text is sent to the model, audio is generated, and nothing is stored — not the text, not the audio, not the request itself. This is the same ZDR commitment you already trust for chat, image generation, and every other feature.

No training on your data. No logs retained by the inference provider. The audio is generated, delivered to your browser, and that is the end of it.

What it costs

AI Voice uses your plan's included AI budget, just like text and image generation. The voice model is priced at $1.00 per million text tokens in and $20.00 per million audio tokens out.

In practice, that means:

• A short paragraph costs fractions of a cent
• A full page of text costs roughly $0.03
• Replaying a message you have already generated is free — the audio is cached in your browser for the session

You can see your budget usage any time in Settings > Billing. And if you are in Settings previewing voices, those previews count toward your budget too — though each one is just a few cents.

The Voice settings page in ARMES shows you a transparent cost breakdown for the active model, so you always know what you are spending before you tap play.

The model behind it

We are launching with Gemini 3.1 Flash TTS, a neural text-to-speech model from Google that delivers natural, expressive audio across 70+ languages. It outputs studio-quality 24 kHz audio and supports inline audio tags for nuanced delivery.

This is a single launch model. We are actively testing additional voice models and plan to add more options soon — all held to the same zero-data-retention standard. As new models are added, you will see them in the Voice Model picker in Settings, with pricing and voice options for each.

You can always see the full, up-to-date model roster and pricing on the AI Access and Pricing pages.

Available on Pro and Ultra

AI Voice is available on Pro and Ultra plans. If you are on the Free or Eco plan, you will see the speaker button but tapping it will invite you to upgrade — and the Voice section in Settings shows exactly what you are unlocking.

Upgrade to Pro to unlock AI Voice →

Or if you are not on ARMES yet — start free, no credit card required. Explore everything else on the Free plan, and upgrade when you are ready to hear your AI think out loud.

Your AI. Your voice. Your terms.