Mimick for the web
Read any PDF out loud in a natural voice, slow or fast, with the words lit as they are spoken — in a browser tab,
with nothing installed. Your document is opened by your own browser and never leaves your computer.
What this does
Read aloud in a natural voice
A Piper voice, running in this tab. Norman comes with Mimick; more voices download once, when you choose them, and work offline after that.
Follow along
As Mimick reads, the sentence and the word being said are lit on the page.
More than PDFs
Word (.docx), OpenDocument (.odt), Rich Text (.rtf), EPUB and FictionBook books, PowerPoint (.pptx) and Impress (.odp) presentations, web pages, Markdown and plain text files open too, laid out as pages — a presentation becomes one page per slide. Save a copy (Ctrl+S) then keeps any of them as a PDF.
A scanned PDF with no text has its words recognised as it opens, here in the browser — a long one asks first, with a button in the bottom bar (or Reading ▾ → Recognise text). A scan with only some of its words — the rest still a picture — is left as it is until you choose Reading ▾ → Recognise text, which adds the words it lacked.
Find your way
A document's table of contents sits beside it (F9), and Ctrl+F finds any text.
Say it right
Teach the voice how to say a name or a term: select it, right-click, How to say.
Read at your speed
From 0.75× to 4×, changed at any moment.
Highlight and annotate
Four colours, notes beside the page in bold, italics or lists, and sticky notes with titles, tied to the passage they are about. They are kept in this browser, and
Save a copy gives you the PDF with them in it, as regular PDF annotations other readers can see, and
Export notes gives you them as a document of their own — Word, OpenDocument, PDF or Markdown.
Take it with you
Convert a document, some pages of it or a passage to an MP3.
Keeps your place
Come back to a document and pick up where you left off, at the same page and zoom.
File ▾ lists the files you opened last, and Forget this document removes every trace of one.
Works offline
Once it has loaded fully, Mimick opens and reads with no internet connection.
How to use it
Open a PDF, a document (.docx, .odt, .rtf), a book (EPUB, .fb2), a presentation (.pptx, .odp), a web page, or a .md or .txt file with File ▾, Ctrl+O, or by dropping it on the page. Press Space to read and pause —
Norman reads to start with; Select a new voice… in the voice box hears and downloads others (about 60–75 MB each, once). Change the speed in the top bar. Select text and press Enter to read just that, or right-click → Start reading from here. Turn on Reading ▾ → Click to read and a click on any sentence reads from there.
Ctrl+F finds text. Ctrl+H highlights, Ctrl+M adds a note, Ctrl+S saves a copy with
them in it. Every key is under Help ▾ → Keyboard shortcuts.
Privacy
Every document you open stays on this computer: your own browser reads it, and nothing is uploaded.
Mimick downloads only the page itself, with Norman, the first time, and each other voice you choose, once, from Hugging Face —
checked against a fixed hash. Your highlights, notes, reading positions and settings are kept in this
browser's storage and nowhere else. No accounts, no analytics, no tracking.
Browsers
Built and tested in Google Chrome, for laptops and desktops. Brave, Edge and other Chromium browsers should
work the same. Firefox has been tried and reads aloud; Safari is untested. In both, files download when they are
finished rather than asking where to save first, there is no Open recent, and Mimick cannot be installed as an app.
Built on
MuPDF (pages, text and annotations) ·
Pyodide (Python in the browser) ·
Piper and
espeak-ng (the voice) ·
ONNX Runtime (running it) ·
LAME, as lamejs (MP3 files) ·
Tesseract, as
tesseract.js (reading scans)
Voices
Every voice is a Piper voice,
trained on recordings people shared, and their licences ask for these credits. Kathleen and Southern English
are built on Ryan, whose licence is non-commercial: they may only be used free of charge, as Mimick is.
| Voice | Recordings | Trained on top of |
|---|
| Norman |
LibriVox readings, public domain; voice by
Bryce Beattie |
— |
| Kathleen |
Kathleen dataset, CC0 |
Ryan: RyanSpeech,
CC BY-NC-SA 4.0 |
| Southern English |
Crowdsourced UK and Ireland English Dialect
speech (OpenSLR 83, Google),
CC BY-SA 4.0 |
Ryan: RyanSpeech, CC BY-NC-SA 4.0 |
| Joe |
OHF-Voice voice datasets, CC0 |
Lessac (Blizzard Challenge 2013, Lessac Technologies) |
| Kusal |
Mycroft AI Mimic 2 recordings |
Lessac |
| VCTK |
CSTR VCTK Corpus,
University of Edinburgh, CC BY 4.0 |
Lessac |
| LibriTTS |
LibriTTS-R (OpenSLR 141, Google),
CC BY 4.0 |
Lessac |
Written by
Claude Code
Arranged by
Kat Hollander · github.com/kathollander
License
Please remix, fix, and break as you please! We're built on open source software, so we're giving it back as open source software.
GNU Affero General Public License, version 3 ·
the source code