marl / marl/openl3

Make linear frontend consistent with mel

Open
#77 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

bug
Dominant language
Jupyter Notebook
Stars
604
Forks
66
PR merge metrics
No merged PRs in 30d

Description

The linear frequency spectrogram frontend results in 197 frames instead of 199, which seems to be caused by:
- In `_librosa_linear_frontend` (`openl3/core.py`), `center=False` should be `center=True` when calling `librosa.stft`
- In `_construct_linear_audio_network` (`openl3/models.py`), `pad_end=True` should be added to argument for __fix_kapre_spec
- In `_construct_linear_audio_network` (`openl3/models.py`), in the `else` block corresponding to `if include_frontend`, the `input_shape` should account for centering

In addition to fixing these, we'll need to regenerate the regression data likely

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start in _librosa_linear_frontend in openl3/core.py and _construct_linear_audio_network in openl3/models.py, checking the listed STFT, padding, and input-shape changes. Then regenerate the regression data and verify that the linear frontend produces 199 frames consistently with mel.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
audio-video-rtc
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Clearly specified
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.