ViuAI_TTS_200M / inference.py

Commit History

29 days ago

  1. Fix: Peak audio normalization (-1 dB), robust speaker search, and tuned CFG for crystal-clear benchmark generation

    Verified signature
    d62a3fe
  2. Fix: Peak audio normalization (-1 dB), robust speaker search, and tuned CFG for crystal-clear benchmark generation

    Verified signature
    0965f29
  3. Fix: Peak audio normalization (-1 dB), robust speaker search, and tuned CFG for crystal-clear benchmark generation

    Verified signature
    4af529a

30 days ago

  1. Master Fix: Real F0 pitch, acoustic Mel-80 to 100 filterbank vocoder projection, soundfile multi-format loader, safe weights_only

    Verified signature
    4d6d5dc
  2. Master Fix: Real F0 pitch, acoustic Mel-80 to 100 filterbank vocoder projection, soundfile multi-format loader, safe weights_only

    Verified signature
    5fda8fd
  3. Master SOTA Upgrade: in-context voice conditioning, decoupled studio vocoder, EMA tracking, warmup scheduler

    Verified signature
    79dd8dc

  1. Update: training fixes, LICENSE + requirements + scripts

    Verified signature
    9c8f79c

  1. Initial commit: ViuAI_TTS_200M architecture, prosody engine, and weights

    Verified signature
    4fe4e70