Skip to content

VOICE_V4 repeated-onset smoothing acceptance

Status: Accepted for VOICE_V4.

VOICE_V4 keeps the VOICE_V3 phrase-continuity renderer and smooths connected token openings that still sounded click-like on repeated subwords.

Scope

  • dootdoot --version reports dootdoot VOICE_V4.
  • The embedded semantic mapping remains the locked token-to-axis table in assets/dootdoot_asset_v1.doot.
  • Connected syllables no longer replay the full attack transient.
  • Connected pitch and vowel openings blend from the previous rendered state.
  • Connected envelope starts ramp through the early body instead of replaying the full attack peak.
  • The committed golden WAV hashes were regenerated under VOICE_V4.

Directional Check

Fixture phrase: hahahahahahahahahahah.

The check measures median derivative roughness in the first 12 ms after connected subword starts relative to a nearby 12 ms body window. Lower is smoother.

setmedian connected-onset roughness ratio
VOICE_V3 dootdoot3.570
VOICE_V4 dootdoot1.821
acceptance regression limit2.400

This is a directional listening aid. The sample-level contract remains the golden WAV hash fixture.

Independent open-source droid acoustics. Not affiliated with Lucasfilm or Disney.