All pull requests

Pull requests

• fix: harden reference API input (ID trailing newline, upload extension)

  #1365·rudycelekli opened Oct 24, 2026·

  ·

• fix: don't crash on a long inline --reference_text (OSError ENAMETOOLONG)

  #1364·rudycelekli opened Oct 24, 2026·

  ·

• fix: correct S2-Pro speaker-tag chunking (dropped leading text; turn vs speaker cap)

  #1363·rudycelekli opened Oct 24, 2026·

  ·

• fix: don't emit a second generic error after a worker error in inference()

  #1362·rudycelekli opened Oct 24, 2026·

  ·

• fix(server): honour an explicit --device instead of forcing CPU

  #1361·li-lizhe opened Oct 20, 2026Contributor·

  ·

• fix(server): dynamic autocast context based on model device instead of hardcoded cuda

  #1360·xuilen opened Oct 19, 2026·

  ·

• Uv

  #1359·lambdawalker opened Oct 18, 2026·

  ·

• Add DGX Spark CUDA 13 support and default uv CUDA stack

  #1348·lambdawalker opened Oct 8, 2026·

  ·
  1

• feat: add CPU reference encoding and persistent token caching

  #1347·clxiny opened Oct 4, 2026·

  ·

• Punjabi Normalizer - converting Gurmukhi text to Latin diacritics for better pronunciation

  #1340·zish-s opened Sep 28, 2026·

  ·
  2

• perf: don't force MATH SDPA backend outside torch.compile

  stale
  #1337·Ray0907 opened Sep 15, 2026Contributor·

  ·
  2

• [pre-commit.ci] pre-commit autoupdate

  #1332·pre-commit-ci[bot] opened Sep 7, 2026Bot·

  ·

• fix: persistent_workers crash at num_workers=0 + warn on silent token truncation

  stale
  #1329·rohitsynth1208 opened Aug 27, 2026·

  ·
  1

• feat: add FunASR dataset annotation tool

  #1309·LauraGPT opened Jul 24, 2026·

  ·
  1

  3

• fix: make pyaudio an optional [client] extra so headless installs work

  stale
  #1304·MushiSenpai opened Jun 28, 2026·

  ·
  5

• fix: reject empty TTS text at the schema boundary (Fixes #946)

  stale
  #1303·MushiSenpai opened Jun 27, 2026·

  ·
  1

  5