All pull requests

Pull requests

• fix(squeezeformer): import Union/Optional from typing, not torch.nn.modules.conv

  #2822·dayn9t opened Oct 16, 2026·

  ·

• [runtime] upgrade openfst to v1.8.5 and WeTextProcessing to v1.2.0

  #2817·pengzhendong opened Jun 15, 2026Member·

  ·
  1

• Fix WeTextProcessing build and add ITN (Inverse Text Normalization) support for iOS runtime

  #2809·Ma-Dan opened Mar 24, 2026Contributor·

  ·

• [bugfix] firered conv2dsubsampling4 and transformer cmvn for padded inputs

  #2806·shen9712 opened Jan 21, 2026·

  ·

• [model] support qwen3omni_aut

  #2801·cdliang11 opened Dec 16, 2025Collaborator·

  ·
  4

• fix firered/subsampling.py where x_mask shape may not be same as x shape

  #2789·yuql-sea opened Oct 24, 2025·

  ·

• fix the bug [TypeError: cannot pickle 'generator' object]

  #2760·MXuer opened Aug 6, 2025·

  ·

• refactor moe of positionwise_feed_forward to let us convert onnx success

  #2726·computervisionlearner opened Apr 26, 2025·

  ·
  6

• ChunkFormer

  #2723·khanld opened Apr 17, 2025·6 of 7 tasks·

  ·
  2

• streaming batch chunk decoding method is available.

  #2704·AdilAdam opened Mar 3, 2025·

  ·
  1

• iOS, fix build issues with new XCode and wetextprocessing

  #2638·Ma-Dan opened Oct 9, 2024Contributor·

  ·

• [Audio LLM]support audiollm for asr, based on whisper and llama3

  #2532·Zth9730 opened May 20, 2024Contributor·

  ·
  6

• Using module kaldi_native_fbank, add the CPU feature extraction option.

  #2496·76176235 opened Apr 19, 2024·

  ·
  1

• [transformer] Add moe_noisy_gate

  #2495·llleohk opened Apr 18, 2024·

  ·
  20

• [tts] add vqtts

  #2325·robin1001 opened Jan 25, 2024Collaborator·

  ·

• [vits] add vits/valle/megabyte support

  #2272·robin1001 opened Jan 4, 2024Collaborator·

  ·
  1

• [wenet] nn context biasing

  #1982·kaixunhuang0 opened Aug 31, 2023Collaborator·

  ·
  109

• Add 'run_batch' mode for GPU encoding and decoding with batch_size >= 1

  #1534·veelion opened Nov 3, 2022Contributor·

  ·
  2

• blockformer

  #1504·LeonWlw opened Oct 19, 2022·

  ·
  5

• self-supervised pretraining(wav2vec 2.0/data2vec) for wenet

  #1003·Emiyassstar opened Mar 31, 2022Collaborator·

  ·
  13