Issues

• Streaming nemotron backend: cache-aware encoder state appears corrupted between consecutive utterances (VAD mode), producing empty final with no preceding partial

  CrispStrobe/CrispASR#547 · Filter by author jostoz opened on Oct 17, 2026

• [Feature-request] Drama 3 tts

  CrispStrobe/CrispASR#546 · Filter by author fablevi opened on Oct 15, 2026
  1

• Heads-up for CrispASR devs: Fish Audio just dropped Drama-3 — a 35B MoE TTS model that’s a natural counterpart to CrispASR.

  CrispStrobe/CrispASR#545 · Filter by author marwanlhabti5-coder opened on Oct 14, 2026
  1

• canary translate crash

  CrispStrobe/CrispASR#532 · Filter by author letsjustfixit opened on Oct 10, 2026
  2

• omnivoice: port the missing upstream output/reference audio post-processing (fade_and_pad_audio, remove_silence, add_punctuation)

  CrispStrobe/CrispASR#518 · Filter by author Bonenk opened on Oct 8, 2026
  1

• New model: Index-echo

  CrispStrobe/CrispASR#485 · Filter by author Oplay66 opened on Oct 1, 2026
  2

• [Improvement] CrispASR should be built with CUDA 12.6 instead of 12.8

  CrispStrobe/CrispASR#483 · Filter by author vyncere opened on Oct 1, 2026
  3

• voxcpm2 RTF is much higher over vulkan

  CrispStrobe/CrispASR#461 · Filter by author huzhua opened on Sep 24, 2026
  19

• New model released : Integrate nyra-forced-aligner for conversational English word timestamps

  CrispStrobe/CrispASR#456 · Filter by author marwanlhabti5-coder opened on Sep 23, 2026
  1