Voice notes and video capture

Say what you know about a part and keep it as your words with Muse Voice Transcribe; send a short video with a capture. Picking the sharpest label frame out of that video is not built.

Built so far

Voice note. On an item's page, under Prepare to sell or give away, signed in, pick a WAV recording (ten minutes at most, within the upload bound of MAX_UPLOAD_MB: about six minutes of 16 kHz 16-bit mono, twice that at 8 kHz). Muse Voice Transcribe turns it into text, kept on the item as the field condition_notes with the source voice_note, accepted as your own words; the draft reads it from there. A transcript longer than the field holds (4000 characters) is cut at a word and the answer says so; say the rest in another note. POST /api/items/{id}/voice. The phone's Talk tab and a capture's spoken note are kept the same way (Android app, Capture from the phone).

Video. A capture from the phone, or from the Inbox's form, carries one short MP4 or 3GP video, kept and played back in the Inbox; no frame is read from it yet, so what the label says has to be in a photo (Next steps).