starred/audiobookshelf
1
0
Fork 0
mirror of https://github.com/advplyr/audiobookshelf.git synced 2026-05-30 23:40:40 +02:00
Code Issues 1.0k Packages Projects Releases 10 Wiki Activity

[Enhancement]: Add de-noising functionality #3048

New Issue
Open
opened 2026-04-25 00:13:15 +02:00 by adam · 1 comment
adam commented 2026-04-25 00:13:15 +02:00
Owner
Copy Link
Copy Source

Originally created by @ak-42 on GitHub (Oct 21, 2025).

Type of Enhancement

Server Backend

Describe the Feature/Enhancement

Books sometimes have a recording noise, such as tape noise, or other issues. This is more common with non-professionally produced books, although I have encountered noise even on Audible-produced audiobooks.

It would be helpful if Audiobookshelf could de-noise those books with a simple interface.

Why would this be helpful?

Noise is distracting, and impacts the listening experience. At the same time, removing the noise, once you know how, is in my experience relatively easy, and works well. Thus, helping users fix up books that have noise is a useful feature for Audiobookshelf to implement.

Future Implementation (Screenshot)

Existing dependency, ffmpeg, provides excellent de-denoising capabilities. I have personally denoised at least 20 books with various types of noise using the following filter and it has worked exceptionally well: -af "arnndn=m=/path/to/models/rnnoise-models/somnolent-hogwash-2018-09-01/sh.rnnn,afftdn=nr=20"

This is a two stage approach of using rnnoise model from https://github.com/GregorR/rnnoise-models. The model is trained on human speech, and in my experience removes 80% of noise. It is followed by the afftdn noise filter which in my experience removes the remainder of the noise. If there is any remaining, I am unable to perceive it. I am also not able to discern any impact on the quality of the non-noise recording.

The model itself is sub-300 kB, and according to the upstream it is not subject to copyright, so it should not be burdensome to include it.

Audiobookshelf Server Version

v2.30.0

Current Implementation (Screenshot)

N/A

Originally created by @ak-42 on GitHub (Oct 21, 2025). ### Type of Enhancement Server Backend ### Describe the Feature/Enhancement Books sometimes have a recording noise, such as tape noise, or other issues. This is more common with non-professionally produced books, although I have encountered noise even on Audible-produced audiobooks. It would be helpful if Audiobookshelf could de-noise those books with a simple interface. ### Why would this be helpful? Noise is distracting, and impacts the listening experience. At the same time, removing the noise, once you know how, is in my experience relatively easy, and works well. Thus, helping users fix up books that have noise is a useful feature for Audiobookshelf to implement. ### Future Implementation (Screenshot) Existing dependency, `ffmpeg`, provides excellent de-denoising capabilities. I have personally denoised at least 20 books with various types of noise using the following filter and it has worked exceptionally well: `-af "arnndn=m=/path/to/models/rnnoise-models/somnolent-hogwash-2018-09-01/sh.rnnn,afftdn=nr=20"` This is a two stage approach of using rnnoise model from https://github.com/GregorR/rnnoise-models. The model is trained on human speech, and in my experience removes 80% of noise. It is followed by the afftdn noise filter which in my experience removes the remainder of the noise. If there is any remaining, I am unable to perceive it. I am also not able to discern any impact on the quality of the non-noise recording. The model itself is sub-300 kB, and according to the upstream it is not subject to copyright, so it should not be burdensome to include it. ### Audiobookshelf Server Version v2.30.0 ### Current Implementation (Screenshot) N/A
adam added the enhancement label 2026-04-25 00:13:15 +02:00
adam commented 2026-04-25 00:13:16 +02:00
Author
Owner
Copy Link
Copy Source

@ak-42 commented on GitHub (Oct 23, 2025):

To further support the case, below is a clip of the US president Franklin Delano Roosevelt (recording in public domain), in the noisy original.
https://github.com/user-attachments/assets/30ab3e7d-7855-44ca-a549-6581099841c4

Same clip, after the denoising via -af "arnndn=m=/path/to/models/rnnoise-models/somnolent-hogwash-2018-09-01/sh.rnnn,afftdn=nr=20".
https://github.com/user-attachments/assets/f6af7f1a-e0ed-4074-a51e-50c9e7c8fc67
(the artifacting that remains is also present in the original, it's just more noticable once the noise is gone)

@ak-42 commented on GitHub (Oct 23, 2025): To further support the case, below is a clip of the US president Franklin Delano Roosevelt (recording in public domain), in the noisy original. https://github.com/user-attachments/assets/30ab3e7d-7855-44ca-a549-6581099841c4 Same clip, after the denoising via `-af "arnndn=m=/path/to/models/rnnoise-models/somnolent-hogwash-2018-09-01/sh.rnnn,afftdn=nr=20"`. https://github.com/user-attachments/assets/f6af7f1a-e0ed-4074-a51e-50c9e7c8fc67 (the artifacting that remains is also present in the original, it's just more noticable once the noise is gone)
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
authentication
backlog
bug
chapter editor
config-issue
ebooks
encoding/embedding
enhancement
help wanted
listening sessions & progress
planned
possible plugin
progress sync
pull-request
Mirrored from GitHub Pull Request
 sorting/filtering/searching
unable to reproduce
upload
users & permissions
waiting
No Label enhancement
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
adam (Adam Melkus)
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: starred/audiobookshelf#3048
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?