Obsidian Voice to Text: Dictate Notes on Mac, Windows and Phone

Obsidian has no voice-to-text button of its own, but you can already talk into any note for free: your computer's or phone's built-in dictation types wherever the cursor is, and Obsidian is no exception. On a Mac, choose Edit → Start Dictation. On Windows, press Windows key + H. On iPhone or Android, tap the microphone on the keyboard.

For more than that, add a plugin:

• Speech Kit, free and offline, for live dictation on desktop with speech models that run on your computer.
• Whisper, free to install, for record-then-transcribe on desktop and mobile with your own OpenAI, Groq or other Whisper-compatible API key.
• SystemSculpt, free to install, records in a note or in AI chat and transcribes on its hosted service with credits, no API key.

We make SystemSculpt. The SystemSculpt steps below match version 6.18.10, checked November 3, 2026; other plugin facts are from each plugin's README and the Obsidian plugin directory on the same day.

Which route fits

Built-in dictation is the quickest start. Pick a plugin when you want to keep the audio, dictate long sessions, or get the same tool on every device.

Use your computer's or phone's dictation

These need no plugin. Put your cursor in a note, start dictation and talk.

Mac. Turn it on in System Settings → Keyboard → Dictation. Then, in a note, choose Edit → Start Dictation, press the Microphone key if your keyboard has one, or use the Dictation shortcut set in the same settings. Press Escape to stop. Apple's text under the Dictation setting says whether your speech is processed on your Mac or sent to Apple.

Windows 10 and 11. Press Windows key + H to open voice typing. Microsoft says it uses online speech recognition, so you need an internet connection.

iPhone, iPad and Android. Tap in a note, then tap the microphone on the on-screen keyboard. Your keyboard does the work, so it behaves in Obsidian the way it does in any other app.

Built-in dictation keeps no recording. If you might need to check what was said, record the audio too.

Record audio without a transcript

Obsidian's core Audio recorder plugin records from your microphone and embeds the recording at the end of the open note. Turn it on in Settings → Core plugins, then select Start/stop recording in the ribbon. It doesn't turn speech into text; use it to keep the audio and transcribe it later with one of the plugins below.

Speech Kit: free, offline dictation on desktop

Speech Kit, formerly Local Dictation, runs speech models on your computer, so dictation needs no account, API key or internet once a model is installed.

1. Install Speech Kit from Settings → Community plugins → Browse.
2. Follow its setup wizard, which installs the speech engine and your first model.
3. Select Try dictation now, or start dictation from the ribbon, command palette or a hotkey.

Words appear while you speak, and the finished text lands as Markdown at your cursor. It can also transcribe local audio and video files. It runs on macOS with Apple silicon, Windows and Linux, and not on Obsidian mobile.

Whisper: record, then transcribe with your own key

The Whisper plugin records, sends the audio to the speech-to-text service you choose, and puts the text at your cursor. It works on desktop and mobile.

1. Install Whisper from Settings → Community plugins → Browse.
2. Add your API key in its settings: OpenAI, Groq, Azure or another Whisper-compatible service.
3. Open a note, press Alt + Q, speak, then press Alt + Q again.

You pay your provider for each minute of audio. Its optional post-processing can send the text to Claude, GPT or a local model such as Ollama to fix grammar, remove filler words or pull out action items.

SystemSculpt: voice notes and dictated chat messages

SystemSculpt records inside Obsidian and transcribes on its hosted service, so there is no key to set up. It also puts a microphone in its AI chat.

Dictate into a note:

1. Install SystemSculpt free from Settings → Community plugins → Browse, and sign in under Settings → SystemSculpt AI → Account.
2. Put your cursor in a note, then run Toggle audio recorder from the command palette, or select Audio Recorder in the ribbon.
3. Speak, then select Stop recording. Wait for Recording saved before you switch away from Obsidian.

With the default Workflow settings, transcription starts when the recording ends and the text goes in at your cursor. Leave the cursor where it was until the text appears: if you move it or switch notes, SystemSculpt leaves the note alone and saves the transcript as its own file. It keeps the audio in your vault too. Turn on Clean up transcript in Settings → SystemSculpt AI → Workflow to fix punctuation and remove filler words.

Dictate a chat message: open SystemSculpt chat and select Record message, the microphone next to the message box. The transcript lands in the message box for you to edit, or turn on Send after dictation to send it straight away. Dictation uses SystemSculpt's hosted transcription even when you chat with your own Claude Code or Codex.

Keep Obsidian open and in front while you record, on desktop or mobile; recording in the background isn't guaranteed. Recording is free, and transcription uses credits. Eligible new accounts can claim 500 one-time starter credits with no card; Monthly ($9 a month) and Yearly ($90 a year) include a monthly credit allowance. See pricing.

For audio files, subtitles, meetings and YouTube videos, see the Obsidian transcription guide.

Privacy

• Speech Kit keeps audio on your computer.
• Built-in dictation follows your device's settings. Windows voice typing sends speech to Microsoft's online service; on a Mac or phone, check your dictation settings.
• Whisper sends audio to the provider whose key you add.
• SystemSculpt sends the audio you transcribe to its hosted service; see the Privacy Policy.

Questions

Does Obsidian have voice to text? Not built in. Obsidian's core Audio recorder records audio but doesn't transcribe it. Your device's own dictation works in any note, and plugins such as Speech Kit, Whisper and SystemSculpt add speech to text.

What is the best speech-to-text plugin for Obsidian? Speech Kit if you want it free, offline and private on desktop. Whisper if you want it on your phone too and already have an OpenAI or Groq key. SystemSculpt if you want recordings, transcripts and AI chat in one plugin with no key.

Can I dictate in Obsidian offline? Yes, with Speech Kit once a model is installed, on Mac, Windows or Linux. Mac and phone dictation can work offline on some devices and languages. Windows voice typing, Whisper with a cloud key and SystemSculpt need internet.

Does voice to text work in Obsidian on iPhone and Android? Yes. The keyboard microphone works in any note for free. Whisper and SystemSculpt also record and transcribe in Obsidian mobile; Speech Kit is desktop only.

Is it free? Built-in dictation, the core Audio recorder and Speech Kit are free. Whisper is free to install, and you pay your provider for the audio you transcribe. SystemSculpt is free to install; transcription uses credits, starting with 500 one-time starter credits for eligible new accounts.

Can Obsidian turn a voice note into a summary? Yes, with a plugin. Whisper's post-processing can tidy a transcript or pull out action items. SystemSculpt's Audio Processor turns a recording into a note with a summary, decisions and action items; see AI meeting notes in Obsidian.

Install SystemSculpt free from the Obsidian plugin directory, or compare its plans for hosted transcription.

Sources

Checked November 3, 2026.

• Apple: Use Dictation on Mac
• Microsoft: Use voice typing to talk instead of type on your PC
• Obsidian: Audio recorder
• Speech Kit: README
• Whisper: README
• SystemSculpt: audio transcription docs, pricing, privacy policy