Written Tutorial
Prefer to read? The same walkthrough, step by step.
What you will learn
Dictation turns speech into text on your own PC: from a recording you already have, live from a microphone, or live from your phone in a meeting room. The video starts with a real community call, 30 minutes with six speakers. Foundry had it written down in 3 minutes and 39 seconds, each speaker in their own colour.
1. Transcribe a recording
Open Foundry and click the Dictation tab. The page in the middle is your document. Click File, then Transcribe a file, and pick an MP3, WAV, M4A or even a video.
It needs no setup. When you want more control, Dictation settings has Hot Words for names Foundry should know, and you can fix the language instead of leaving it on Automatic. On the Professional plan, the Multiple Speakers switch tells who said what.
2. Who said what, and checking the text
With Multiple Speakers on, every speaker gets their own colour. Click Speakers, pick one and give them their real name.
Want to check a sentence? Click any word and press Play from cursor, and you hear exactly that moment. Follow text scrolls the page along while it plays. To change a word, just type: it's a normal document.
3. Your phone as the microphone
Next to the Dictate button, the input button chooses what Foundry listens to: your PC's microphone, the sound playing on your PC, or your phone. Choose Foundry Link, then Connect phone, and a code appears. Scan it with your phone's camera. A page opens in the phone's browser, so there's no app to install.
Keep that page open while you record (it's your phone's line to your PC). If it closes by mistake, open the same page again and the phone reconnects. Once you press Disconnect on your PC, the page stops working for good, and next time you scan a new code at your PC. So connect your phone before you leave home.
On the phone, tap Microphone, switch Mic on, lay the phone on the table and tap Start dictation. Everything said in the room appears on your PC as text while it is being said. Watch the line under the document: the blue words at its end are still being listened to, and Foundry may still change them, because the next words often show what was really said. When they turn white, they're settled, and the line moves up into your document.
Your phone and your PC don't need the same Wi-Fi. The phone's sound travels to your PC over an encrypted connection, every connection gets its own new address, and the AI work happens on your PC. Your phone can be in a café or a lecture hall while Foundry writes at home. When you're done, tap Stop dictation and switch the microphone off.
4. Write by talking
Press Dictate or F8 and speak. Foundry types what you say into the document and puts in the commas and full stops by itself. For a new paragraph, say "command new paragraph". Press Dictate again to stop.
If you pause a lot while you think, choose Careful in Dictation settings. Foundry then waits longer before it finishes a sentence.
5. Sound playing on your PC
Foundry can also write down what plays on your PC. Choose System audio as the input, play a webinar or a video lesson, and press Dictate.
6. Save and send
Save keeps it all, the text, the recording and who said what, and the document opens again exactly as you left it. To use the text somewhere else, choose Save as and pick RTF, which opens in Word, or plain text. Export makes an MP3 of the recording with the text as a file next to it, ready to send.
7. Your AI agent
On the Professional plan, your own AI agent can do this for you. Tell it to transcribe every recording in a folder, and it hands each one to Foundry and brings back the texts. Foundry comes with written instructions for agents in its documentation\agent_skills folder.
Four tips
Voice memos work too. Copy one from your phone to your PC, choose Transcribe a file, and your idea becomes text.
Finding a quote. Looking for one sentence in an hour-long interview? Click near where it was said and listen, instead of scrubbing through the audio.
A quiet voice goes missing. Open the input settings next to Dictate and set Sensitivity to More.
Subtitles. Use the Dubbing tab: it transcribes a video and exports an SRT file.
Your recordings stay on your PC
The AI runs on your own graphics card. Your recordings and transcripts are saved on your PC, they never reach our servers, and nobody at Demodokos reads them.
Recordings in the video
Community call: Open Source Collective, Community Call, 16 September 2021, CC BY 4.0. The first 30 minutes, transcribed in Foundry as shown. Meeting room: AMI Meeting Corpus, meeting ES2002a, University of Edinburgh and the AMI consortium, CC BY 4.0. An excerpt, sent to Foundry through the Foundry Link phone page.