Psychomotor change is one of the most reliable observable signs in depression, and half of it happens in the timing of speech rather than its content. A depressed speaker takes longer to begin answering, leaves longer silences inside a turn, articulates more slowly, says less, and flattens the melody of the voice. Szabadi's 1976 finding — that pause time alone tracks motor retardation and responds to treatment — has been replicated with acoustic tooling for fifty years, and it is the single strongest case for annotating recordings instead of transcripts.
This is the most agent-assistable scale in the audio library. Four of the five signs are arithmetic over a time-stamped transcript: the gap between the end of the interviewer's question and the start of the reply, the silence inside a turn, words per minute of phonation, and total words per answer. Joey can propose those and show you the numbers it used. The fifth — vocal monotony — needs the ear, and its curated skill instructs the agent to abstain rather than guess at pitch from a transcript.
The thresholds below are operational conventions, not published cut-offs. The literature reports group differences and effect sizes, not diagnostic boundaries, and every threshold moves with language, interview style, telephone versus in-person, and the interviewer's own pace. Treat them as a calibration starting point for your own study, record any change in your codebook, and check inter-rater agreement before trusting cross-study comparisons.
Rate the subject, never the interviewer. A slow interviewer produces long gaps that have nothing to do with the patient. And rate against the same speaker's baseline where you have one: the clinically meaningful signal in a treatment study is change over time, not a single absolute value.
Related scales: SANS Alogia (text) covers poverty of speech and increased latency as clinician judgements from the interview content; TLC (text) covers pressure and poverty of speech as language phenomena. Use this scale when you want the timing measured off the recording.