May 20

1. 13:28

   fix(pipeline): wire TextEncoder style_key into VectorEstimator.uncond_masker (mirror gitea 052b24d)

   ambassadia
   Verified signature

   8970e7a
2. 10:25

   feat: streaming + custom voice mixing API

   ambassadia
   Verified signature

   cf6c0c7
3. 10:07

   quality: VE style_key fix — closes 500-800 Hz centroid gap on dark voices

   ambassadia
   Verified signature

   60f9c22
4. 09:36

   stability: bump default seed to 99 + re-gen samples

   ambassadia
   Verified signature

   20372df
5. 09:16

   fix(critical): add missing residual in DurationPredictor.sentence_encoder

   ambassadia
   Verified signature

   277aeb1
6. 09:15

   fix: DP residual + re-gen samples (Whisper validation 86% overlap)

   ambassadia
   Verified signature

   e57ba25
7. 08:00

   security: strip absolute paths revealing dev machine + private monorepo

   ambassadia
   Verified signature

   bcfacb5
8. 07:11

   v0.1.0 — initial release (MLX-native Supertonic 3, ~x100 RTF on M4)

   ambassadia
   Verified signature

   69dee76