Button time and recording time are not the same
The moment a person presses record is not always identical to the moment the system begins capturing sound. The screen may respond immediately while microphone permission, input switching, or buffer preparation follows. If the click under the finger is presented as the start signal, a person can reasonably believe the opening sentence was captured in full. A recording tone should follow the time of the recording, not simply the time of the button.
That makes placement the first decision, before timbre. The whole product needs one rule: either speak after the signal, or know that recording is already active when the signal begins. When sound, status graphics, and haptics point to different moments, people end up checking the waveform because they no longer trust the boundary. A good recording tone is not a decorative celebration of a successful tap. It is the line after which words will remain.
The start should open; the stop should close
Playing one sample forward for start and backward for stop makes the pair feel related, but the distinction can collapse in a quiet room, inside a pocket, or through a small speaker. Do not give pitch direction the entire semantic burden. Attack, duration, and the weight of the final sound should also describe the action. A start tone makes room for what comes next; a stop tone places a period after what will no longer be captured.
The two cues can share material while using different grammar. A start might place a small sense of forward motion between two compact elements and leave its ending lightly open. A stop can use the same core timbre with a faster decay and firmer cadence. The point is not a formula such as “up means start, down means stop.” The test is whether one hearing makes the body understand which cue permits the next action and which one closes the state.
Be heard in the room, then leave the file alone
Unlike most alerts, a recording tone is likely to leak into the artifact it creates. A long tail, diffuse high frequencies, or excessive playback level can cover the first syllable even after the cue has technically ended. A tone that is too thin and quiet, however, disappears in a café or rehearsal room. A short body, quick decay, and a midrange contour that survives a small speaker are useful starting points. The aim is not to leave a beautiful tail; it is to return the ear to the conversation immediately after the boundary is heard.
The final judgment belongs at real listening distances, not only on studio monitors. Hear the device on a table, in a hand, and inside a bag or pocket. The cue should not jump out harshly in a quiet room, and noise should not make start and stop interchangeable. Keep the same boundary visible and tactile for anyone who has disabled sound. Recording handles private time, so the failure of one sensory channel should never blur the edge of consent.