audio

audio copied to clipboard

The TextProcessor object from TTS pipeline, should be able to return Tensors in specified device. Currently the returned Tensors have to be manually moved to the target device. ```python bundle...

mthrok

Updates necessary to WaveRNN training script

2

1. Get rid of `bg_iterator` https://github.com/pytorch/audio/blob/a6f9cf8babfb096381e914e23950371478672b3e/examples/pipeline_wavernn/main.py#L13 This is known to slow down the process, and was removed from the library. 2. Replace `DataParallel` with `DistributedDataParallel`. https://github.com/pytorch/audio/blob/a6f9cf8babfb096381e914e23950371478672b3e/examples/pipeline_wavernn/main.py#L325 `DataParallel` is officially deprecated...

mthrok

Currently WaveRNN's forward method expects client code to to pad the input spectrogram to the specific size (`kernel_size - 1 // 2`). This breaks the encapsulation. The WaveRNN's forward method...

mthrok

ERROR: Could not find a version that satisfies the requirement torchaudio (from versions: none) ERROR: No matching distribution found for torchaudio

3

### 🐛 Describe the bug ERROR: Could not find a version that satisfies the requirement torchaudio>=0.5.0 (from asteroid) (from versions: none) ERROR: No matching distribution found for torchaudio>=0.5.0 (from asteroid)...

clort81

There is a contraction in WaveRNNInferenceWrapper during inference when batched

1

The waveform above is the ground truth and the one below is the generated waveform with `batched` set to true. See this [colab example](https://colab.research.google.com/drive/19-LC9MhNQFsoSH93b4hPWXXcycmoutmb?usp=sharing) ([internal](https://colab.research.google.com/drive/1d-kmdYALKDwGw-u5lTb8JAP7dyKFtWbk?usp=sharing) for more information).

yangarbiter

Remove insignificant test assets

14

@astaff had introduced guideline for test assets in https://github.com/pytorch/audio/pull/759 and we can get rid of the following existing assets. - [x] `100Hz_44100Hz_16bit_05sec.wav` sine wave, should be replaced by on-the-fly generation....

mthrok

kaldi.fbank alternative in librosa?

1

## ❓ Questions and Help Hi everyone, I would really appreciate if someone could let me know how to replicate **compliance.kaldi.fbank()** function in **librosa**? I've gone through alot of literature...

AlexJian1086

cuda_version can't be changed in CI. It needs to be improved.

1

## 🐛 Bug In unittest_windows_gpu, there's an environment variable of CUDA_VERSION. https://github.com/pytorch/audio/blob/ecd068f583e4e45ef80635d975ac2bf38d4e819a/.circleci/config.yml.in#L514-L520 But the value couldn't be changed in fact, because it will be revert to 10.2 by regenerate.py https://github.com/pytorch/audio/blob/ecd068f583e4e45ef80635d975ac2bf38d4e819a/.circleci/regenerate.py#L212...

mszhanyi

Loading a BytesIO opus file does not seem to work

4

## 🐛 Bug Loading a BytesIO opus file does not seem to work. ## To Reproduce Steps to reproduce the behavior: ``` import torchaudio import io print(torchaudio.__version__) # samples from...

christopherhesse

Add support for bitrate in sox_io backend

3

`soxi` shows the files bitrate, but I'm pretty sure `torchaudio.info` doesn't when using the `sox_io` backend. ## 🚀 Feature Create functionality within `torchaudio.info` and the `AudioMetadata` object to be able...

rbracco