-
Notifications
You must be signed in to change notification settings - Fork 3.6k
Pull requests: NVIDIA-NeMo/Speech
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Derive word-confidence boundaries from the decoded text itself
ASR
community-request
#16064
opened Aug 13, 2026 by
Eoin-Houstoun
Loading…
Coalesce missing batch_size in ClusteringDiarizer dataloader configs
ASR
community-request
#16063
opened Aug 13, 2026 by
Eoin-Houstoun
Loading…
Remove PR babysitter automation, update commit/PR check instructions
CI
#16062
opened Aug 13, 2026 by
pzelasko
Collaborator
Loading…
Preserve valid codec subframe before audio EOS
TTS
#16059
opened Aug 13, 2026 by
vklimkov-nvidia
Member
Loading…
Fix variable name for audio file path in '_write_to_tar' function
community-request
#16058
opened Aug 13, 2026 by
advait-bm
Loading…
5 of 8 tasks
Add native THD packed ASR encoders
ASR
common
#16053
opened Aug 11, 2026 by
pzelasko
Collaborator
•
3/3
Loading…
Replace pointwise convolutions with matmul in ConformerEncoder: 10% E2E training speedup
ASR
community-request
waiting-on-maintainers
Waiting on maintainers to respond
#16052
opened Aug 11, 2026 by
MahmoudAshraf97
Contributor
Loading…
2 of 8 tasks
feat(speechlm2): add optional MTP training modes
#16050
opened Aug 10, 2026 by
Slyne
Contributor
Loading…
Support chunk microbatching and idxpack for multimodal SpeechLM data
common
#16048
opened Aug 10, 2026 by
KunalDhawan
Collaborator
•
2/3
Loading…
[Streaming SpeechLM] Add audomodel support
ASR
#16045
opened Aug 7, 2026 by
stevehuang52
Collaborator
•
Draft
add simulstream backend + qwen reasoning models for ST
ASR
#16043
opened Aug 7, 2026 by
lilithgrigoryan
Collaborator
Loading…
8 tasks
Flash RNN-T Loss: 2x Faster Loss and up to 50x VRAM Reduction
ASR
community-request
waiting-on-maintainers
Waiting on maintainers to respond
#16040
opened Aug 6, 2026 by
MahmoudAshraf97
Contributor
Loading…
5 of 8 tasks
Adding MoE, PEE, GGEMM then SALM, vLLM related changes
ASR
Run CICD
#16038
opened Aug 6, 2026 by
tango4j
Collaborator
•
1/3
Loading…
5 of 8 tasks
Add an AutoModel Path for EMTTS In Addition to Local NemotronH file
TTS
#16036
opened Aug 5, 2026 by
blisc
Collaborator
Loading…
1 of 3 tasks
Enable offloading of validation models to CPU during training runs for EMTTS
TTS
#16035
opened Aug 5, 2026 by
blisc
Collaborator
Loading…
1 of 3 tasks
fix: report manifest format mismatch instead of failing in the wrong parser
common
community-request
waiting-on-maintainers
Waiting on maintainers to respond
#16033
opened Aug 5, 2026 by
adityasingh2400
Loading…
1 of 4 tasks
Implement Prosody metrics for TTS evaluation
common
TTS
#16031
opened Aug 5, 2026 by
Edresson
Member
Loading…
[TTS] EasymagpieTTS Online CFG Distillation
CI
TTS
#16030
opened Aug 5, 2026 by
artem-gorodetskii
Contributor
•
Draft
Fix multi-GPU deadlock in RNN-T validation (fuse_loss_wer)
ASR
community-request
waiting-on-maintainers
Waiting on maintainers to respond
#16029
opened Aug 5, 2026 by
Sahith59
Loading…
Prevent fused RNN-T WER validation deadlocks
ASR
community-request
waiting-on-maintainers
Waiting on maintainers to respond
#16027
opened Aug 4, 2026 by
SiavashShams
Loading…
3 of 8 tasks
Fix Lhotse training limits for prompt ASR models
ASR
community-request
waiting-on-maintainers
Waiting on maintainers to respond
#16019
opened Aug 4, 2026 by
SiavashShams
Loading…
4 of 8 tasks
Support BF16 activations in the Numba RNN-T losses
ASR
community-request
waiting-on-maintainers
Waiting on maintainers to respond
#15996
opened Jul 30, 2026 by
MahmoudAshraf97
Contributor
Loading…
4 of 8 tasks
Keep the configured RNN-T loss across Waiting on maintainers to respond
change_vocabulary
ASR
community-request
waiting-on-maintainers
#15995
opened Jul 30, 2026 by
MahmoudAshraf97
Contributor
Loading…
3 of 8 tasks
Previous Next
ProTip!
Follow long discussions with comments:>50.