<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Greg Daynes</title>
    <description>The latest articles on DEV Community by Greg Daynes (@greg_daynes_fa28219b855a4).</description>
    <link>https://dev.to/greg_daynes_fa28219b855a4</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4208928%2F5dd803d8-ed2c-40ba-9af7-fdffc13faf98.png</url>
      <title>DEV Community: Greg Daynes</title>
      <link>https://dev.to/greg_daynes_fa28219b855a4</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/greg_daynes_fa28219b855a4"/>
    <language>en</language>
    <item>
      <title>I built a dictation app that never sends your voice to a server</title>
      <dc:creator>Greg Daynes</dc:creator>
      <pubDate>Mon, 26 Oct 2026 15:06:59 +0000</pubDate>
      <link>https://dev.to/greg_daynes_fa28219b855a4/i-built-a-dictation-app-that-never-sends-your-voice-to-a-server-gnn</link>
      <guid>https://dev.to/greg_daynes_fa28219b855a4/i-built-a-dictation-app-that-never-sends-your-voice-to-a-server-gnn</guid>
      <description>&lt;p&gt;Most dictation apps work the same way. You talk, your audio goes up to a cloud server, and text comes back. That's fine for a shopping list. It's less fine when you're dictating a legal draft, patient notes, or an email about a deal that hasn't been announced.&lt;/p&gt;

&lt;p&gt;So I built &lt;strong&gt;Quiet Dictation&lt;/strong&gt;: hold a hotkey, speak, release, and clean, punctuated text appears wherever your cursor is. The difference is that speech recognition runs entirely on your own laptop. No audio is uploaded or stored anywhere else.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why on-device matters
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Privacy:&lt;/strong&gt; your voice never leaves your machine, so there's nothing on a server to leak, subpoena or train on.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Offline:&lt;/strong&gt; it works on a plane, behind a hospital firewall, or on terrible cafe Wi-Fi.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Cost:&lt;/strong&gt; with no servers transcribing audio, it can be cheap. It's $5/month or $39/year. Wispr Flow is $15/month.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  What it does
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Works in any app: Word, Outlook, Gmail, Google Docs, Slack, WhatsApp, your browser&lt;/li&gt;
&lt;li&gt;Double-tap the key for hands-free mode when you're dictating something long&lt;/li&gt;
&lt;li&gt;Personal dictionary for names and terms it should always get right&lt;/li&gt;
&lt;li&gt;Meeting recording and transcription&lt;/li&gt;
&lt;li&gt;Tuned for non-American and non-native English accents. I'm in South Africa, and most dictation tools treat my accent as an edge case.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;It runs on Mac (Apple Silicon, macOS 13+) and Windows 10/11. There's a 14-day free trial, no card needed.&lt;/p&gt;

&lt;p&gt;If you type all day, try it and tell me what breaks: &lt;a href="https://www.quietdictation.com" rel="noopener noreferrer"&gt;quietdictation.com&lt;/a&gt;&lt;/p&gt;

</description>
      <category>productivity</category>
      <category>privacy</category>
      <category>ai</category>
      <category>showdev</category>
    </item>
  </channel>
</rss>
