Replace known non-human audio, such as a voice assistant or music.

The body is the whole list. [] clears it. Every span must lie within the video. 409 video_duration_unknown until the duration is known. 422 non_human_span_out_of_range when a span ends after the video, 422 non_human_spans_clock_unsupported when the video was analysed from start_millis > 0, and 422 non_human_spans_too_large when the stored list exceeds its size cap.

Sending this list does not by itself remove a speaker from analysis results. A span is recorded time. A speaker marked system or media on the audio is what leaves Moments.

Owner or media:write. Not available to a delegated token. No charge.