Create a word cloud from any text or .txt file. Seven shapes, five palettes, stop-word filtering, and PNG or SVG download. Runs entirely in your browser.
This word cloud generator turns a block of text into a shaped, coloured frequency visualisation you can download as a PNG or an SVG. Paste text or upload a .txt file, adjust the filters, and the cloud redraws as you type — larger words appear more often in the source. There is no signup, no watermark, and no upload: the text is counted and the image is drawn entirely in your browser, so survey verbatims, meeting notes, and support tickets never leave your machine.
Word clouds earn their place when the question is “what is this corpus mostly about?” — the first pass over open-ended survey responses, a quick read of what customers complain about, a slide that has to convey a theme in three seconds, or a classroom exercise on frequency distribution. They are a summarising device, not a measuring device; the section on reading them below is worth two minutes before you put one in front of an executive.
Alongside the canvas, a ranked frequency list shows the actual counts behind the picture — useful because a cloud tells you which words are big, while the list tells you by how much. If you want the underlying statistics rather than the visualisation, the word counter reports totals, unique terms, and reading time for the same text.
The pipeline is straightforward and worth understanding because each stage changes what you see. First the text is tokenised into words and lower-cased, so Security and security count as one term. Tokens shorter than the minimum length are dropped, numbers are dropped unless you asked for them, and stop words are removed if the toggle is on. What remains is counted, sorted by frequency, and truncated to your maximum word count.
Font size is then scaled between a minimum and maximum according to each word’s count relative to the most frequent word, so the largest word in the cloud is always the modal term and everything else is sized in proportion. Placement uses a spiral algorithm: the biggest word is positioned first near the centre, and each subsequent word spirals outward from a candidate point until it finds a position that collides with no already-placed word and lies inside the shape mask. The mask is a geometric test — for the circle, a point is accepted when x² + y² ≤ 0.92 in normalised coordinates; the heart, star, and diamond use their own inside-the-shape predicates.
Two consequences follow from that algorithm. Position carries no meaning: a word near the centre is not more important than one at the edge, it was simply placed earlier. And rotation is a packing decision, not a semantic one — a vertical word is vertical because it fit that way.
The visual encoding is area, and human beings are poor at comparing areas. A word twice as frequent does not look twice as big, and a long word at a modest count can occupy more pixels than a short word at a high one — infrastructure at 20 occurrences will out-shout risk at 30. That is the fundamental limitation, and it is why the frequency list matters.
Two further caveats. Word clouds destroy context: not is a stop word, so “not helpful” and “helpful” both contribute to a big cheerful helpful. And they treat inflections as separate terms unless you normalise first — delay, delays, and delayed occupy three slots and split their combined weight three ways, which can push a genuinely dominant theme below the cut-off.
Used with that in mind, a cloud is an excellent opening move: it shows you which terms deserve a real query. Use it to generate hypotheses, then verify them with counts, sentiment analysis, or by reading the source responses. Do not use it as evidence that one theme outweighs another — for that, a bar chart of the same frequency list is more honest and takes no longer to make.
No. Tokenising, counting, layout, and image export all happen in your browser with JavaScript and the HTML canvas. The text and the resulting image never leave your device, which makes the tool safe for confidential survey responses and internal documents.
Plain text only — the picker accepts .txt. For a PDF, Word document, or spreadsheet, copy the text out and paste it into the box.
Stop-word removal is on by default and filters common English function words. Turn the toggle off if you genuinely want them, though the result is usually dominated by grammar rather than meaning.
PNG for slides, documents, and anywhere a fixed-size raster is fine, and it is the only format that carries the transparent background. SVG when the cloud will be printed large, embedded on a page that scales, or recoloured later in a vector editor, since it stays sharp at any size.
Yes — choose Transparent under background style and download the PNG. It will sit on any slide colour without a white box around it.
Enough for frequencies to mean something. A few hundred words produces a thin cloud; a few thousand gives a stable distribution. If yours looks sparse, lower the minimum frequency before adding more text.
No. There is no stemming or lemmatisation, so delay, delays, and delayed are counted separately. If that splits an important theme, normalise the wording in your source text before pasting it in — a search-and-replace pass with the regex tester is the quickest way to work out the pattern first.
No. Placement is decided by a spiral packing algorithm that fits each word wherever there is space inside the shape. Only size carries information, and only relative to the most frequent term. For a precise comparison, read the frequency list next to the canvas.
A word cloud (also called a tag cloud) is a visual representation of text data where the size of each word indicates its frequency or importance within the source text. Word clouds transform raw text into instantly interpretable graphics — readers can grasp the dominant themes and keywords of a document at a glance without reading the entire text.
Word clouds are widely used in content analysis, market research, educational settings, and presentation design. By visualizing word frequency, they reveal patterns, biases, and themes that might be missed when reading text linearly.
The generation process follows these steps:
A word cloud is a visual representation of text data where the most frequently used words appear larger than less common ones. It is useful for quickly identifying themes in customer feedback, survey responses, meeting notes, or any large body of text. The visual format makes patterns immediately recognizable.
You can download your word cloud as a PNG image for presentations and documents, or as an SVG vector file for high-quality printing and further editing in design software. Both formats preserve the colors and layout of your generated visualization exactly as you see it in the preview.
You can adjust several filters: set the maximum number of words (20-180), minimum word length to exclude short words, and minimum frequency to include only words that appear multiple times. You can also toggle stop word removal to filter common words like the, and, and is, and choose whether to include numbers.
The tool offers five color palettes: Aurora (blues and purples with accent colors), Heatmap (reds to greens), Pastel (soft muted tones), Mono (grayscale), and Ocean (deep blues). You can also choose between light, dark, or transparent backgrounds to match your presentation or document style.
Yes, you can upload plain text files up to 5MB in size by clicking the Upload .txt button. The tool will process the file contents and generate a word cloud automatically. This is convenient for analyzing larger documents without copying and pasting.
The Word Stats panel displays the total number of tokens detected, unique words after filtering, number of words filtered out by your settings, and average word length. It also shows a ranked list of your top 10 most frequent words with their occurrence counts.