Press to talk.
Ctrl+Fn
or
Ctrl+Win
Press and hold the dictate hotkey. The hotkey is global, works in any app or browser tab.
Voice-to-text that sounds like you.
polished
We could push the launch to Thursday or Friday, whichever works for the team.
Hold a hotkey, talk, release.
Auris transcribes, polishes, and inserts clean text into your text box.
Bring your own API keys (BYOK), or run fully local and private.
Math
You speak faster than you type.
The speed has always been there.
The cleanup is what was missing. Auris handles both.
The cleanup is what was missing. Auris handles both.
How
Hold. Speak. Release.
One hotkey, any app.
Two seconds from press to polished text.
Two seconds from press to polished text.
Say anything.
Talk for as long as you need
Auris transcribes everything while you're holding the key. No time limit.
Read it back.
Clean text · 1–2s later
Clean text lands in the window you started in. Filler gone, punctuation right, your voice intact.
Voice
Pick your voice.
Three sliders, same words.
Different voice. Drag a slider →
Different voice. Drag a slider →
You said
"hey just wanted to check in see how youve been doing havent talked in a while would love to grab coffee or something soon"
Auris polished
Hi, want to grab coffee soon?
Per-app presets
Different apps, different tones.
Auris detects the active window and switches tone.
Set the preset once per app; forget it forever.
Set the preset once per app; forget it forever.
Slack
modecasual
toneplayful
fmtmd
You say
"Yeah we should um ship this on friday i think"
Auris types
We should ship this Friday.
Gmail
modeemail
toneprofessional
fmtplain
You say
"hi sarah um can we move the meeting"
Auris types
Hi Sarah,
Can we move the meeting?
Can we move the meeting?
Notion
modedefault
toneeditorial
fmtmd
You say
"todo migrate the api"
Auris types
## TODO
- Migrate the API
- Migrate the API
Cursor
modecode
tonedefault
fmtplain
You say
"function get user by id takes a string returns a user"
Auris types
function getUserById(id: string): User
ChatGPT
modeminimal
tonedefault
fmtplain
You say
"explain this um step by step with examples"
Auris types
Explain this step by step, with examples.
VS Code
modecode
tonedefault
fmtplain
You say
"todo refactor the auth middleware"
Auris types
// TODO: refactor the auth middleware
Defaults
Cleanup work, built in.
Stray em-dashes, fillers, missing punctuation, wrong caps.
Auris fixes them by default and keeps your text humanized.
Auris fixes them by default and keeps your text humanized.
Raw
"well —, actually —, yes"
Polish
"well, actually, yes"
Em-dashes gone.
Most LLMs love em-dashes. Most people want them gone. Auris strips them and uses commas instead.
Raw
hello she said its mine
Polish
“Hello,” she said. “It’s mine.”
Smart quotes.
Curly quotes and apostrophes. Auris also adds them around dialogue when you don't say "quote."
Raw
um well uh i think like we should ship this you know
Polish
Well, I think we should ship this.
Fillers stripped.
Um, uh, like, mm, you know, basically. Dead syllables removed automatically. Tunable in Settings.
Raw
hi sarah. can we move the meeting
Polish
Hi Sarah. Can we move the meeting?
Capitals where they belong.
First letter after every period, question mark, exclamation. Proper nouns too. Auris reads context like a copy editor.
Raw
i need this by friday what do you think
Polish
I need this by Friday. What do you think?
Punctuation that wasn't there.
Commas, periods, question marks added where the sentence demands. Run-ons split. Lists formatted.
Raw
tasks sprint planning auth fix
Polish
## Tasks
- Sprint planning, auth fix
- Sprint planning, auth fix
Structure when you need it.
Enumerate and Auris turns it into bullets. Say "heading" for a heading. Shift topic for a paragraph break.
Run it
Three ways to run Auris.
Auris doesn't host an AI.
You bring your own API keys, or run one on your own machine.
You bring your own API keys, or run one on your own machine.
01 Cloud
Bring your API keys.
Plug in your OpenAI, Anthropic, or Groq API keys. Auris talks to them directly.
STTCloud
PolishCloud
02 Local
Fully local.
Whisper for speech, any local LLM for polish, or built-in rules. Audio stays local and private.
STTLocal
PolishLocal
03 Mixed
Mix and match.
Dictation and polish are independent. Swap them as you want.
STTLocal or Cloud
PolishLocal or Cloud
Backend
Use any AI, or go fully local.
Bring your own API keys (BYOK).
Or run a local model and stay private. No internet needed.
Or run a local model and stay private. No internet needed.
BYOK · Default
Anthropic
Built around Claude Haiku. Fast, cheap, and recommended.
BYOK
OpenAI
Plug in your key for any GPT model, from gpt-4o-mini to gpt-4o.
BYOK
Groq
LPU hardware delivers sub-second polish. Generous free tier to start.
Local
On your machine
Run a local model. Fully offline and private.
0
Privacy · in local mode
data leaves your machine.
Run Whisper locally for transcription, Ollama (or any local LLM) for polish. Auris's outbound network traffic is empty. No telemetry, no analytics, no error reporting, no update pings, no model-training payloads. Pull the network cable. It still works.
Audio runs locally or goes direct to your provider. Nothing goes to us.
No email, no password, no profile.
Buy the license and it's yours.
$49 once. No recurring charges. Updates included.
Your dictation never trains a model. Yours, used by you, then gone.
What does exist
~/Library/Auris/history.db
One local file.
Guarantees
What we stand behind.
Four claims, four proofs.
No marketing fluff, no buzzwords.
No marketing fluff, no buzzwords.
On-Device Audio
Run local for transcription and audio that never leaves your machine. Data is opened, processed, discarded in RAM.
Run itNo Telemetry
Zero analytics, zero crash reports, zero usage stats. No opt-out switch because there's nothing to opt out of.
PrivacyBYOK or Local
Bring your Anthropic, OpenAI, or Groq key. Or run a local LLM for full privacy. Auris never proxies your keys.
BackendOpen Formats
Memos as .md, history as .jsonl, dictionary as .json. Nothing locked inside a proprietary archive.
WorkspaceVersus
$49 once.
Not $15 every month.
Wispr Flow charges $15/mo, forever. Auris is $49 once.
Same job. Different math.
Same job. Different math.
Feature
Auris
Competitors
One-time / lifetime
$49 Early Access · $79 Standard
Wispr: subscription onlySuperwhisper: $249
Monthly
$2 to $4 in API credits$0 running fully local.
Wispr: $15/moSuperwhisper: $8.49/mo + APIs
Local mode
Included. Zero cloud calls.
Wispr: cloud onlySuperwhisper: included
Tone control
Three sliders, per-app modes
Wispr: basic customizationSuperwhisper: basic customization
Money-back
30 days, no questions
Wispr: 14 daysSuperwhisper: unstated
License
One price.
Every feature.
Early Access pricing
Early Access
$49once
$30 off Standard pricing.
Full app, every feature.
After Early Access
Standard
$79once
After early access pricing ends.
$30 more, same app.
.edu email required
Student
$29once
Verified .edu email.
Full app, every feature.
3 to 50 seats
Team Self-Hosted
$59per seat
Volume discounts at 10, 25, 50+ seats.
FAQ
The usual questions.
N.01
Will there be a free version or a free trial?
Yes, both, at public launch.
The free version runs the core dictation flow (hotkey, transcribe, insert) with a limited feature set. Use it as long as you want, no time limit, no card required. If you outgrow it, upgrade once and you're done.
The free trial unlocks the full app: every feature, every model, every preset, for a fixed window after install. When the trial ends, you decide: buy a license, or drop back to the free version. Your settings and dictionary stay.
Either path, no subscription. You pay once or you don't pay at all.
N.02
Do I need to set anything up after I buy?
Two minutes. The first time you open Auris, a setup walks you through giving microphone access and picking where Auris should process your speech: your own API key at OpenAI, Anthropic, or Groq, or kept entirely on your machine. Pick one, click through, start talking.
N.03
What does the API usage actually cost?
Most people pay $2 to $4 a month.
Go local instead and the only cost is electricity.
N.04
What if I don't like it?
30-day money-back, no questions.
Reply with the word "refund" and we'll process it within 24 hours.
N.05
Will it run on my computer?
If you're on Mac or Windows from this decade, yes.
Linux is on the way.
For cloud speech-to-text (STT) and cloud polish, any modern laptop is fine. The work happens on the API.
For local Whisper or local LLM polish, your CPU and GPU do the work. Apple M1 and newer handle it comfortably. Older Intel Macs and lower-end Windows machines will run it, but transcription and polish will be slower. If your hardware can't keep up, run the cloud backends for the fastest results.
N.06
How accurate is it?
Accurate enough that most people stop noticing they're dictating.
If a word it doesn't know shows up wrong, fix it once and Auris remembers.
Stop typing. Start talking.
Auris is currently in beta. Shipping soon.
30-day money-back, no questions.
Mac & Windows · Lifetime updates