Skip to content

Velora 0.10.17

Choose a tag to compare

View all tags
@sushilk1991 sushilk1991 released this 24 Jul 11:28
v0.10.17
bb54b2f

What changed

  • Isolates the on-device writing model in a persistent child process so a native MLX stall cannot hold the final result indefinitely.
  • Returns the raw transcript at the existing cleanup deadline, kills only the stalled writing worker, and warms a replacement automatically.
  • Preserves the existing Whisper/Qwen models, prompt cache, formatting behavior, streaming path, modes, language handling, and raw-history fallback.
  • Adds separate model-cleanup, cleanup-wall, and stop-to-final measurements to local Voice Intelligence.

Quality benchmark: 23/23 exact-model cases passed with every prompt-cache hit.

SHA-256: ce7237d45332755e1d07ead8cae08482f5598cad5bd823c96a63407d46e7eeb9

Assets 3
Loading