Skip to content

Add RMSNormalization transform for target-level RMS normalisation (closes #176)#183

Open
Shohail-Ismail wants to merge 4 commits intoiver56:mainfrom
Shohail-Ismail:feature/rms-normalisation
Open

Add RMSNormalization transform for target-level RMS normalisation (closes #176)#183
Shohail-Ismail wants to merge 4 commits intoiver56:mainfrom
Shohail-Ismail:feature/rms-normalisation

Conversation

@Shohail-Ismail
Copy link

@Shohail-Ismail Shohail-Ismail commented Jun 27, 2025

Summary

Adds a new RMSNormalization transform (subclass of BaseWaveformTransform) that:

  • Converts a target RMS level in dBFS into a linear gain factor
  • Computes the combined‐signal RMS, applies gain = target_amp / (rms + eps)
  • Returns results in an ObjectDict (samples, sample_rate, etc.)

Tests

Includes three basic tests covering:

  1. Adjusting a 440 Hz sine wave from -20 dBFS to -10 dBFS
  2. Leaving silent input unchanged
  3. Applying equal gain to both channels of a stereo signal

Issue

Closes issue #176

Let me know if you’d like any further tweaks, thanks!

@iver56 iver56 self-requested a review June 28, 2025 06:56
iver56
iver56 requested changes Jun 28, 2025
View reviewed changes
Copy link
Owner

@iver56 iver56 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the PR!

For consistency with the other transforms, I would like RMSNormalization to also have/support mode and p_mode arguments.

Also, let p default to 0.5 for now, so it has the same default as the others. I realize that this default might not be ideal, but consistency is good. An overall change in the default p in the future will need to be a separate effort.

@Shohail-Ismail
Copy link
Author

Shohail-Ismail commented Jun 28, 2025 •
edited
Loading

Thanks for the feedback! Args mode and p_mode are now supported, and p defaults to 0.5 Let me know if there is anything further.

@Shohail-Ismail Shohail-Ismail requested a review from iver56 July 2, 2025 20:06
@iver56
Copy link
Owner

iver56 commented Jul 2, 2025

Thanks for the update! I will have another look soon. I'm on vacation these days. That is what has delayed the review. Thanks for the patience

@Shohail-Ismail
Copy link
Author

Shohail-Ismail commented Jul 30, 2025

Hi @iver56, I would appreciate it if you could take a look when you get the chance. Thanks!

iver56
iver56 requested changes Jul 31, 2025
View reviewed changes
Copy link
Owner

@iver56 iver56 left a comment •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I ran pytest. It gave me an error:

torch_audiomentations\augmentations\rms_normalization.py:7: in <module>
    from ..utils import db_to_amplitude
E   ImportError: cannot import name 'db_to_amplitude' from 'torch_audiomentations.utils' (C:\Users\iver5\Code\torch-audiomentations\torch_audiomentations\utils\__init__.py)

@Shohail-Ismail
Copy link
Author

Shohail-Ismail commented Jul 31, 2025 •
edited
Loading

@iver56 Apologies, I'm not sure why this wasn't apparent in my pytest a month ago, but regardless I have made the changes and pytest finishes cleanly.

@iver56
Copy link
Owner

iver56 commented Aug 3, 2025

Thanks! I will have a look

@Shohail-Ismail Shohail-Ismail requested a review from iver56 August 26, 2025 06:46
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@iver56 iver56 Awaiting requested review from iver56

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@Shohail-Ismail @iver56