From 788b9c3325d922ab67037e8594339f60a0f23dce Mon Sep 17 00:00:00 2001 From: Omkar Tripathi <38633467+triggeredcode@users.noreply.github.com> Date: Mon, 12 Oct 2026 16:39:41 +0530 Subject: [PATCH] docs: dictation works with any app that pastes; the free and open-source ones are listed Wispr Flow was named as the way to dictate, because it is what the author started with. The doc (renamed docs/wispr.md to docs/dictation.md) now says that any app that pastes works, and lists OpenWhispr (MIT, macOS, Windows, Linux, runs on the computer), VoiceNative (the author's own, macOS, runs on the computer) and the system's own dictation next to Wispr Flow, with where each runs and what it costs. The README, the guide, the site and the architecture note say "a dictation app". Co-Authored-By: Claude Sonnet 5.5 --- README.md | 7 ++++--- docs/architecture.md | 2 +- docs/dictation.md | 49 ++++++++++++++++++++++++++++++++++++++++++++ docs/guide.html | 18 ++++++++-------- docs/wispr.md | 24 ---------------------- site/index.html | 12 ++++++----- 6 files changed, 70 insertions(+), 42 deletions(-) create mode 100644 docs/dictation.md delete mode 100644 docs/wispr.md diff --git a/README.md b/README.md index 77b2b5c..041d950 100644 --- a/README.md +++ b/README.md @@ -72,8 +72,9 @@ What stays true whatever you say: - **You stay in charge.** Every change by the model is exactly one undo step, any earlier change can be reverted on its own, and your words are never lost: if the model cannot use them, they are kept as written. - **One view.** The page, one input box under it, an optional history panel. It scales to a full-screen window. -- **Voice or keyboard.** Dictate with [Wispr Flow](https://wisprflow.ai) (optional) or just type. Lucid takes the text from - there. +- **Voice or keyboard.** Dictate with any app that pastes text, or just type. [Wispr Flow](https://wisprflow.ai), + [OpenWhispr](https://github.com/OpenWhispr/openwhispr) (free, open source) and + [VoiceNative](https://github.com/triggeredcode/voicenative) (free, macOS) are covered in [docs/dictation.md](docs/dictation.md). - **A toolbox for AI agents.** An MCP server lets Claude Code, or a model in LM Studio's chat, read and edit the same blocks. ## Install @@ -219,7 +220,7 @@ wrong are in [`docs/mcp.md`](docs/mcp.md). Lucid has no account, no analytics and no cloud service. Your notes are files on your disk, and the language model runs in LM Studio on your computer. The only time Lucid's own PDF engine uses the internet is the one-time download of typesetting -files described above. If you dictate with Wispr Flow, that is a separate app with its own privacy terms; Lucid only sees the +files described above. If you dictate with another app, that is a separate app with its own privacy terms; Lucid only sees the text it pastes. ## Build from source diff --git a/docs/architecture.md b/docs/architecture.md index 524a05a..c6ed065 100644 --- a/docs/architecture.md +++ b/docs/architecture.md @@ -6,7 +6,7 @@ the same way. ```mermaid flowchart TD - say["Speech (Wispr Flow) or typing"] --> box["Input box"] + say["Speech (any dictation app) or typing"] --> box["Input box"] box --> input["Input mode
notes → blocks"] box --> command["Command mode
instructions → ops"] box --> guided["Guided builders
diagram and table grammar"] diff --git a/docs/dictation.md b/docs/dictation.md new file mode 100644 index 0000000..9f315d9 --- /dev/null +++ b/docs/dictation.md @@ -0,0 +1,49 @@ +# Dictating into Lucid + +Lucid does no speech recognition. A dictation app turns your voice into text and pastes it, and Lucid's input box takes the +paste. Any app that pastes (or types) into the focused window works; you do not need to pick one of these. + +| App | Where | Cost and source | Speech runs | +| --- | --- | --- | --- | +| [Wispr Flow](https://wisprflow.ai) | macOS, Windows | commercial, closed source | in the cloud | +| [OpenWhispr](https://github.com/OpenWhispr/openwhispr) | macOS, Windows, Linux | free, MIT | on your computer (Whisper or Parakeet models), or your own cloud key | +| [VoiceNative](https://github.com/triggeredcode/voicenative) | macOS | free, made by Lucid's author | on your computer (WhisperKit, Neural Engine) | +| The system's own | macOS Dictation, Windows Voice Typing (Win+H) | built in | depends on the system | + +Lucid was first used with Wispr Flow, because that is what its author had at hand, and the automated tests paste the way Wispr +does. It is a good place to start, but it is one option. If you want something free and open source, or that keeps your voice on +your computer, use OpenWhispr or VoiceNative, which paste the same way. Their privacy terms are theirs, not Lucid's: Lucid only +sees the text that arrives. + +Two things to know: + +- A **paste** is sent about half a second after the last piece arrives, with no Enter. Text that is typed key by key is not + sent on its own: press Enter. +- Lucid's own command key is the **right ⌥ key** (right Alt on Windows and Linux). Choose another key for your dictation app's + push-to-talk (VoiceNative's is right Shift) so they do not meet. + +## Try it with your voice (a check-list) + +The automated tests paste with ⌘V, which is what these apps do; only you can check it with your voice. About ten minutes. +Read ⌘ as Ctrl and ⌥ as Alt on Windows and Linux; on Linux, use OpenWhispr or type. + +1. **Plain dictation.** Open a page, hold your dictation key and say a thought ("Load balancer. Multiple application servers + behind it. Health checks to avoid unhealthy servers."). The text appears in the input box and the notes appear in the page + about half a second later, without pressing Enter. ✅ +2. **A long ramble** (hands-free mode, 30 to 60 s about caching). One set of notes, even if the app pasted it in pieces. ✅ +3. **Focus somewhere else.** Click **History**, **Markdown** or the page text, then dictate. The words still go to the input + box. ✅ (Click into a field of Settings and dictate: the words go into that field, not the box.) +4. **Correct before sending.** Dictate a sentence and, within half a second, type a word. The automatic send is cancelled; + fix the text and press Enter. ✅ +5. **A command by voice.** Hold the **right ⌥ key** and dictate "cross out MongoDB" (with a list on the page). It is run as + a command, one undo step. Say "undo". ✅ Then say "command mode" alone: the mode changes, nothing is written. ✅ +6. **Say "Lucid, add a row for cost to this table"** in Input mode: one command, then back to notes. ✅ +7. **The diagram builder.** Say "diagram mode", then "client, server, DNS, database", then "connections", then "client to + server". Each phrase is read as a step. "Done" inserts the diagram. ✅ +8. **Clipboard.** After a paste press ⌘V yourself somewhere else. Dictation apps often restore your previous clipboard; note + what you get, it is not Lucid's doing. ℹ️ +9. **The app's own command mode** (Wispr Flow's Fn+Ctrl, for example) is not used by Lucid. If you try it, nothing should be + double-processed or crash. + +If a step does not behave as written, please open an issue and say which step, and which app (and "style", if it has one: it +can change the text before pasting) you use. diff --git a/docs/guide.html b/docs/guide.html index 3cf2d2b..1b37733 100644 --- a/docs/guide.html +++ b/docs/guide.html @@ -141,7 +141,7 @@

On this page

Note formats Saving, export and import Export - Dictating with Wispr Flow + Dictating When something goes wrong Where things live For AI agents @@ -160,7 +160,7 @@

In one minute

  1. Have LM Studio running with the model loaded. The small pill at the top of Lucid shows a green dot and the model name when it is ready. If it says No model loaded, click the Load … button in the pill (it asks first, and takes about half a minute).
  2. Click New page (top of the left column), type a title, press Enter.
    3. -
  3. Talk or type into the box at the bottom. With Wispr Flow just dictate: the words appear in the box and are sent about half a second after you stop. Typing: press Enter.
    4. +
  4. Talk or type into the box at the bottom. With a dictation app just dictate: the words appear in the box and are sent about half a second after you stop. Typing: press Enter.
  5. Watch the page. Your words become notes. To change something, say what to change (Command mode, below). To undo anything, say undo.
  6. Download PDF (top right) when it looks right.
@@ -452,14 +452,14 @@

Export

-

Dictating with Wispr Flow

-

Lucid does no speech recognition. Wispr Flow turns your voice into text and pastes it; Lucid's input box takes the paste.

+

Dictating

+

Lucid does no speech recognition. A dictation app (Wispr Flow, OpenWhispr, VoiceNative, the system's own) turns your voice into text and pastes it; Lucid's input box takes the paste.

@@ -477,7 +477,7 @@

When something goes wrong

Tectonic, the PDF engine, was not foundThe copy of Lucid is damaged. Download it again from the project’s releases (Help, Check for Updates…). Markdown export works meanwhile. PDF without some diagrams, and a note that they could not be drawnThe note says which. Send the page to Lucid again, or fix the diagram in its builder (click the diagram), then download the PDF again. Please report it if it keeps happening. Revert is greyed outThe blocks that change touched were changed after it (by another change or by you). Use undo, or fix it in Edit mode. - Nothing happens when you dictateClick once in the Lucid window so it has the focus; check that Wispr Flow is pasting (the text should appear in the box). + Nothing happens when you dictateClick once in the Lucid window so it has the focus; check that your dictation app is pasting (the text should appear in the box). @@ -502,9 +502,9 @@

Honest limits

diff --git a/docs/wispr.md b/docs/wispr.md deleted file mode 100644 index f91c518..0000000 --- a/docs/wispr.md +++ /dev/null @@ -1,24 +0,0 @@ -# Trying Lucid with Wispr Flow (a check-list) - -Lucid does no speech recognition: Wispr Flow turns your voice into text and pastes it, Lucid's input box takes the paste. The -automated tests paste with ⌘V (what Wispr does); only you can check it with your voice. About ten minutes. (Wispr Flow runs on -macOS and Windows; on Windows read ⌘ as Ctrl and ⌥ as Alt. It has no Linux version: there you type.) - -1. **Plain dictation.** Open a page, hold Wispr's key and say a thought ("Load balancer. Multiple application servers behind - it. Health checks to avoid unhealthy servers."). The text appears in the input box and the notes appear in the page about - half a second later, without pressing Enter. ✅ -2. **A long ramble** (hands-free mode, 30 to 60 s about caching). One set of notes, even if Wispr pasted it in pieces. ✅ -3. **Focus somewhere else.** Click **History**, **Markdown** or the page text, then dictate. The words still go to the input - box. ✅ (Click into a field of Settings and dictate: the words go into that field, not the box.) -4. **Correct before sending.** Dictate a sentence and, within half a second, type a word. The automatic send is cancelled; - fix the text and press Enter. ✅ -5. **A command by voice.** Hold the **right ⌥ key** and dictate "cross out MongoDB" (with a list on the page). It is run as - a command, one undo step. Say "undo". ✅ Then say "command mode" alone: the mode changes, nothing is written. ✅ -6. **Say "Lucid, add a row for cost to this table"** in Input mode: one command, then back to notes. ✅ -7. **The diagram builder.** Say "diagram mode", then "client, server, DNS, database", then "connections", then "client to - server". Each phrase is read as a step. "Done" inserts the diagram. ✅ -8. **Clipboard.** After a Wispr paste press ⌘V yourself somewhere else. Wispr normally restores your previous clipboard; note - what you get, it is not Lucid's doing. ℹ️ -9. **Wispr's own command mode** (Fn+Ctrl) is not used by Lucid. If you try it, nothing should be double-processed or crash. - -If a step does not behave as written, please open an issue and say which step, and which Wispr "style" you use (it can change the text before pasting). diff --git a/site/index.html b/site/index.html index e7f036f..62239ef 100644 --- a/site/index.html +++ b/site/index.html @@ -315,8 +315,9 @@

Talk, refine, export. You stay in charge.

  • Say or type

    - One input box. Dictate with Wispr Flow on macOS or Windows (optional), or just type. - Your words are never lost: if the model cannot use them, they are kept as written. + One input box. Dictate with any app that pastes text (Wispr Flow, OpenWhispr, + VoiceNative), or just type. Your words are never lost: if the model cannot use them, + they are kept as written.

  • @@ -946,11 +947,12 @@

    Good to know.

    - Do I need Wispr Flow? + Do I need a dictation app?

    - No. Dictation is optional: Wispr Flow pastes your words and Lucid takes the paste. - You can just type. Wispr Flow is a separate app with its own privacy terms. + No. Dictation is optional: a dictation app pastes your words and Lucid takes the + paste. You can just type. Wispr Flow, OpenWhispr (free, open source) and + VoiceNative (free, macOS) are separate apps with their own privacy terms.