Issues

• Parakeet TDT: CUDA illegal memory access in the CUDA-graph greedy decoder after GPU memory is released between transcribe() calls

  bugcommunity-request
  NVIDIA-NeMo/Speech#16325 · Filter by author jbrink42 opened on Sep 30, 2026

• ImportError: cannot import name 'spacing' from 'pangu' — pangu is unpinned and pangu>=5.0.0 removed spacing (breaks import nemo.collections.llm on a fresh 2.4.1 install)

  community-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16317 · Filter by author dmvevents opened on Sep 29, 2026

• Cache-aware streaming skips a final chunk shorter than sampling_frames, so a stream's last word depends on its batch

  community-requestwaiting-on-customer
  NVIDIA-NeMo/Speech#16310 · Filter by author kzos opened on Sep 28, 2026
  1

• Cache-aware streaming pipeline runs out of slots when streams end with delete_state() instead of an is_last request

  community-request
  NVIDIA-NeMo/Speech#16309 · Filter by author kzos opened on Sep 28, 2026
  2

• [SPEECHLM2] : vllm serve crashes for dense Nemotron backbones

  bugcommunity-request
  NVIDIA-NeMo/Speech#16289 · Filter by author AudranBert opened on Sep 21, 2026

• nemotron-labs-voicechat-11b: Training data

  community-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16286 · Filter by author arianaa30 opened on Sep 18, 2026

• Sortformer splits one speaker into two when their distance to the mic changes

  community-requestwaiting-on-customer
  NVIDIA-NeMo/Speech#16278 · Filter by author johnSHLee96 opened on Sep 17, 2026
  1

• nemotron-labs-voicechat: Issues with the model's voice

  bugcommunity-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16252 · Filter by author arianaa30 opened on Sep 14, 2026

• Early-interruption augmentation can drop the relocated EOS token entirely

  community-request
  NVIDIA-NeMo/Speech#16247 · Filter by author udsy19 opened on Sep 12, 2026

• ConvSubsampling forward broken: chunking=-1, striding_conv1d/dw_striding_conv1d, and vggnet all raise TypeError

  community-request
  NVIDIA-NeMo/Speech#16223 · Filter by author ManoharPaturi opened on Sep 7, 2026

• SimpleSegmentedTextAggregator never treats a sentence ending in a decimal number as complete

  community-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16218 · Filter by author udsy19 opened on Sep 5, 2026

• FastPitchModel_SSL / SSLDisentangler crash on validation: on_validation_epoch_end() missing 1 required positional argument: 'outputs'

  community-request
  NVIDIA-NeMo/Speech#16216 · Filter by author udsy19 opened on Sep 5, 2026

• [Bug] NeMo Docker image works on single GPU but not on multi-GPU setup

  bugcommunity-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16211 · Filter by author sauravsingla opened on Sep 3, 2026

• [ALERT] Access to NVIDIA's Self-Hosted Runners is Expiring

  nvidia-runners
  NVIDIA-NeMo/Speech#16207 · Filter by author nvidia-runner-mgmt opened on Sep 2, 2026

• IPLEpochStopper ignores enable_stop, always requests a training stop

  community-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16182 · Filter by author udsy19 opened on Sep 1, 2026

• AudioAugmenter loudness-norm disablement leaks across batch rows in add_room_ir_to_batch / add_mic_ir_to_batch

  community-request
  NVIDIA-NeMo/Speech#16180 · Filter by author udsy19 opened on Sep 1, 2026

• voice_agent: NemoSTTService's default model is misspelled ('nnvidia'), silently defaulting has_turn_taking to False

  community-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16178 · Filter by author udsy19 opened on Sep 1, 2026

• verify_speakers_batch returns an unindexable 0-d array for a single audio pair

  community-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16172 · Filter by author udsy19 opened on Aug 30, 2026

• [TTS] AlignerModel bin_loss warmup ramp (bin_loss_scale) has no effect

  community-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16170 · Filter by author udsy19 opened on Aug 30, 2026

• prepare_text_and_asr_labels silently drops EOS when delay_text_channel_by shifts it into the delay window

  community-request
  NVIDIA-NeMo/Speech#16164 · Filter by author udsy19 opened on Aug 29, 2026

• malsd_batch beam search decodes labels for a batch row with encoder_output_length == 0

  community-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16161 · Filter by author udsy19 opened on Aug 28, 2026

• Verify evals on Papers with Code

  community-requestwaiting-on-customer
  NVIDIA-NeMo/Speech#16156 · Filter by author NielsRogge opened on Aug 28, 2026
  1

  

• Abnormal Punctuation and Excessive Sentence Segmentation in German (canary-180m-flash)

  bugcommunity-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16154 · Filter by author Ranger2000 opened on Aug 27, 2026
  3

• EncDecHybridRNNTCTCBPEEOUModel.validation_pass raises UnboundLocalError: text_pred on its default path

  community-requestwaiting-on-maintainers
  NVIDIA-NeMo/Speech#16150 · Filter by author udsy19 opened on Aug 27, 2026

• Relax the lightning<=2.4.0 constraint for NeMo Speech ASR

  NVIDIA-NeMo/Speech#16144 · Filter by author sarahyurick opened on Aug 26, 2026