F0 / spectral-tilt coherence

The Coherence tile on the dashboard answers a simple question: "are the primary pitch cue and the spectral-tilt voice-quality cue telling the same story?" It is a descriptive correlation cue between two metrics that are each already on the dashboard in their own right, not a new perceptual claim and not a verdict.

What it compares

Vocal Analyzer takes two percentile positions within the active goal profile:

• Pitch percentile — where your median F0 sits inside the profile's pitch band (from 0 at the masculine floor to 1 at the feminine ceiling).
• Spectral-tilt percentile — where your preferred voice-quality value sits inside the profile's spectral-tilt band (same 0–1 scale).

The coherence score is a measure of how aligned those two percentile positions are. The tile surfaces three classifications:

• Matched — both cues point the same way within the profile. Typical for recordings where F0 and voice quality both fall into, say, the feminine half of their respective bands.
• Neutral — one or both cues sit near the middle of their band. The classification is not informative because the cues themselves are not strongly one-sided.
• Divergent — the two cues point in opposite directions. For example, a feminine-percentile F0 with a masculine-percentile spectral tilt.

How to read it

• Matched does not mean "you sound feminine" (or masculine). It means the two cues are consistent within the active profile's overlay. Whether listeners perceive the voice consistently is a separate question that coherence cannot answer.
• Divergent does not mean "something is wrong." It means the F0 percentile and the tilt percentile sit on opposite sides of the profile's midpoint. That is a known, normal acoustic state — many speakers land there.
• Neutral is the common state in the androgynous overlap region and should not be over-interpreted.

Caveats

• Coherence depends on the active goal profile (see Goal profiles). Switching profiles changes which bands percentile positions are computed against, which changes the classification even though nothing about your recording changed.
• Coherence is skipped when either input metric is at low or unavailable confidence. A coherence badge of "unavailable" means one of the underlying percentile computations did not have reliable source data.
• Coherence is descriptive. It does not assert that "cue agreement" predicts listener judgement — the perceptual literature on multi- cue integration is more nuanced than that (see Voice and gender perception).

References

• Leung, Y., Oates, J., & Chan, S. P. (2018). Voice, articulation, and prosody contribute to listener perceptions of speaker gender. Journal of Speech, Language, and Hearing Research 61(2), 266–280.
• Pisanski, K., & Rendall, D. (2011). The prioritization of voice fundamental frequency or formants in listeners' assessments of speaker size, masculinity, and attractiveness. The Journal of the Acoustical Society of America 129(4), 2201–2212.