Start with an exported audio file
Save the recording from your phone or recorder as a file you can select on your computer. An iPhone voice memo may be exported as M4A; the actual audio codec still determines whether your browser can read it. Open that file in the workspace above. Transcribana does not connect to Apple, WhatsApp or a voice-note account. It cannot read an app link or recover an inaccessible message. For other inputs, see the audio-format guidance.
Keep short notes in their spoken language
Select the language if you know it, then start transcription. Auto Detect examines the opening audio; a very short note, a musical introduction or a name said in another language may make that sample less representative. English, French and Arabic remain in their original language. Use a current desktop Chrome browser for the best-supported workflow; a phone recording does not imply that heavy transcription will run reliably on that phone.
Check the details that make a note useful
Listen for names, addresses, dates and quantities before editing. For example, “send fifteen copies on Thursday” needs more careful checking than its punctuation. Keep uncertainty visible in your working notes rather than filling a gap from memory. Local writing suggestions are optional and do not verify the recording. Do not accept a correction that changes a deliberate spoken phrase. This is a transcript, not an automatically organized task list.
Save the text before leaving
Copy Text or download TXT from the working version to keep your manual edits. SRT, VTT and JSON retain the original timed recognition segments, so they will not inherit corrections made in the readable document. Download before reloading or navigating away; the page is not a cloud notebook. Model and runtime downloads still use the network even though the memo stays local. Read what local processing does and does not mean before opening a sensitive recording.