Export Documents by Voice: Say the Order, Get the Invoice

How it works

The order box sits at the top of the generator, with a "Say your order" button (a microphone) next to "Make the documents". Tap it and allow the microphone if the browser asks. While you speak, the words appear in the box. After a pause of about two seconds, or when you tap the microphone again, the text is sent as if you had typed it, and the documents fill in.

You can then say or type the next line to add or change something: "The buyer is Nordic Home AB, Malmo, Sweden", "make it 3,000 mugs", "payment T/T 30 days". Your own company details stay as they are, and Undo goes back one step.

The same microphone is in the order box at the end of every tool, template and guide page on the site (it opens the generator with the documents filled in) and in the AI assistant's shipment mode, which answers with the documents and the checks for the destination.

What to say so the documents fill in correctly

Say what an invoice needs, in one go. For goods entering the United States, for example, the invoice rules list the parties, a detailed description of the goods, the quantities, the purchase price of each item, the currency and the country of origin (19 CFR 141.86). Other countries ask for much the same.

• The product in plain words: what it is and what it is made of ("ceramic mugs", not a model code alone)
• The quantity and its unit ("2,400 pieces", "120 cartons")
• The unit price with the currency ("USD 1.85 each")
• The Incoterms rule with the named place ("FOB Busan", "CIF Rotterdam")
• The buyer's company, city and country
• If you know them: payment terms, cartons and weights, the port of loading

Names, numbers and checking

Speech recognition usually writes numbers as digits and is good with common words, but it can mishear company names, similar-sounding numbers (fifteen and fifty) and product codes. Say prices the way you normally would; the assistant reads the amount and the currency. After the documents fill in, read the buyer's name, the quantities and the prices on the preview. To correct something, say or type the line again ("make it 1,500") or edit the field.

Product descriptions and ports come back in English on the documents, whichever language you speak.

Privacy: what goes where

The speech is turned into text by the browser's own speech recognition, not by Triplicate. In some browsers, such as Chrome, the audio is sent to the browser maker's web service for this; Safari uses Apple's speech recognition. Triplicate receives only the text, as when you type, and sends it to its order assistant (Cloudflare Workers AI) to fill in the form. The text is not stored, and the documents are made and saved in your browser.

When the microphone is missing or does not work

The microphone appears only where the browser has speech recognition: Chrome, Edge, Safari and Samsung Internet. Firefox has none, and browsers inside apps (a link opened in a messenger, for example) usually have none; open the page in the phone's browser instead. If the browser says the microphone is blocked, allow it for triplicatedocs.com in the browser's site settings. If nothing was heard for eight seconds, listening stops; tap the microphone and speak again.