Issues

• forced_align returns a sub-optimal path when two predecessor states tie

  pytorch/audio#4221 · Filter by author happyarts opened on Aug 23, 2026

• transforms.MFCC: top_db clamp reduces over the time axis and cannot be disabled

  pytorch/audio#4220 · Filter by author isaacsight opened on Aug 21, 2026
  3

• transforms.Vad silently returns an empty tensor when the waveform contains one non-finite sample

  pytorch/audio#4216 · Filter by author Kayvan-Zahiri opened on Aug 10, 2026
  1

• rnnt_loss: per-sequence logits offset is computed in int and overflows for large-vocabulary batches

  pytorch/audio#4215 · Filter by author happyarts opened on Aug 3, 2026
  1

• torchaudio.save silently saturates PCM-scale int16 input after the TorchCodec migration

  pytorch/audio#4211 · Filter by author John6666cat opened on Jul 30, 2026
  1

• forced_align (CPU) segfaults when frames * (2*targets+1) exceeds 2^31: 32-bit loop index overflow

  pytorch/audio#4208 · Filter by author happyarts opened on Jul 22, 2026
  1

• Conformer output depends on padding, silently hurting accuracy at inference (convolution module ignores lengths)

  pytorch/audio#4205 · Filter by author proshian opened on Jul 12, 2026
  1

• Python 3.14 support

  pytorch/audio#4202 · Filter by author Clayblockunova opened on Jun 19, 2026

• Could you please add the cu132 variant to support Torch 2.12.0?

  pytorch/audio#4194 · Filter by author slavonnet opened on May 15, 2026
  1

• How to use torchaudio without ffmpeg?

  pytorch/audio#4182 · Filter by author JohnHerry opened on Mar 5, 2026
  1

• Pointer arithmetic bug in first_matrix__bitonic_topk_kernel causes shared memory out-of-bounds write

  pytorch/audio#4181 · Filter by author hjldegit opened on Feb 26, 2026
  2

• Compatibility issues with torchaudio on NVIDIA Blackwell (GB10) / aarch64 using NGC PyTorch 25.12

  pytorch/audio#4169 · Filter by author chenqy2 opened on Jan 16, 2026
  1

• Is the Speech Command Classification tutorial still available? If not, why has it been removed?

  pytorch/audio#4168 · Filter by author jchwenger opened on Jan 11, 2026

• ITokenizer type hint error in wav2vec2 aligner pipeline

  pytorch/audio#4153 · Filter by author danijel3 opened on Dec 11, 2025 · 2.11
  1

• How can I use one streamwriter to write multiple videos?

  pytorch/audio#4132 · Filter by author Z-NAVY opened on Nov 17, 2025

• ffmpeg libraries now must be installed, breaking on users machines

  pytorch/audio#4123 · Filter by author jonahclarsen opened on Oct 28, 2025
  3

• Packaging Regression in v2.9.0: torchaudio.load fails on non-shared Python builds

  pytorch/audio#4121 · Filter by author Manamama opened on Oct 20, 2025
  4

• Wav2Vec2Model can't be exported with torch.onnx.export()

  pytorch/audio#4116 · Filter by author Tomas542 opened on Oct 9, 2025
  1

• Add save functionality to CTCDecoder

  pytorch/audio#4113 · Filter by author FredHaa opened on Oct 2, 2025

• Gammatone Filterbanks and Gammatone Spectrogram

  pytorch/audio#4094 · Filter by author samtoone91 opened on Sep 6, 2025

• bandreject_biquad throws error with cuda tensors

  pytorch/audio#3988 · Filter by author asifjalal opened on Jul 23, 2025
  1

• Loading audio file with torchaudio fails (memory crash)

  pytorch/audio#3919 · Filter by author noorbraik opened on May 25, 2025
  1

• io.UnsupportedOperation: seek when using torchaudio.io.StreamWriter with a File-like object

  pytorch/audio#3918 · Filter by author digicosmos86 opened on May 23, 2025
  2

• import_fairseq_model example produces mismatched features and reference output

  pytorch/audio#3914 · Filter by author xefonon opened on May 20, 2025
  1

• UnpicklingError: Weights only load failed. This file can still be loaded occured on document's code of [tacotron2_pipeline_tutorial.html]

  pytorch/audio#3911 · Filter by author f2janyway opened on May 19, 2025