audio

audio copied to clipboard

### 🐛 Describe the bug It is not a proper bug, but it prevents the use of torch script when saving the model. I traced back the issue to the...

giacomopiccinini

torchaudio/lib/libtorchaudio_sox.so, 0x0006): Library not loaded: @rpath/libsox.dylib

5

torchaudio/lib/libtorchaudio_sox.so, 0x0006): Library not loaded: @rpath/libsox.dylib installed pip install sox not work

luohao123

FFmpegBackend failed to load .wav files that contain 30 channels.

3

### 🐛 Describe the bug `FFmpegBackend` failed to load .wav files that contain 30 channels, while `SoxBackend` works well. Code to reproduce: ```python import torchaudio print("load with sox") wav, sr...

ease-zh

streamreader add_video_stream doesn't seem to accept any filter_desc options

1

### 🐛 Describe the bug I'm using the following options in my streamreader: ``` vr.add_video_stream( frames_per_chunk=decode_size, decoder=codec, decoder_option={"threads": "0", "gpu": "0"}, hw_accel='cuda', filter_desc=f"format=pix_fmts=rgb24" ) ``` Unfortunately I get the error...

caspersmit-sa

Hello, i have a problem while installing some of the libraries because i can't install module fcntl. Is there any solution because on one windows pc works but on my...

Toplica001

### 🚀 The feature How would someone go about configuring AV-HuBERT to work with `torchaudio.pipelines.Wav2Vec2FABundle`? It currently only supports [MMS_FA](https://pytorch.org/audio/stable/pipelines.html#pertrained-models) ### Motivation, pitch Currently the `torchaudio.pipelines.Wav2Vec2FABundle` forced aligner only supports...

bejjani

torchaudio need to save segmented chunks as opus like here: torchaudio.save( ..format = 'opus',..

2

### 🚀 The feature I would want to save the file as opus like this: Example: torchaudio.save( os.path.join(args.dst_dir, segment_path), waveform.unsqueeze(0), sample_rate = 16000, format = 'opus',

npovey

The ffmpeg integration shouldn't download linux binaries when the UNIX cmake variable is defined - it should issue proper messages for non-Linux UNIX systems

3

### 🐛 Describe the bug The cmake code assumes [here](https://github.com/pytorch/audio/blob/main/third_party/ffmpeg/multi/CMakeLists.txt#L59) that UNIX implies linux, which isn't correct. The code should check whether OS is Linux and only download linux binaries...

yurivict

Torchaudio.load() reference not found in Pycharm

1

### 🐛 Describe the bug Pycharm Professional 2023.3 is unable to locate the reference to basic functions like torchaudo.save() or torchaudio.load(). I do not not why this happens, maybe the...

asusdisciple

WLS2 compatibilty issue

1

### 🐛 Describe the bug This repo https://github.com/facebookresearch/brainmagick works fine on Ubuntu vanilla. On the exact same WLS2 configuration, the following error arises. This is using torchaudio 2.2.1 ``` Traceback...

arnodelorme